nit

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 1.0,
+    "eval_accuracy": 0.3399546485260771,
+    "eval_loss": 4.18359375,
+    "eval_runtime": 6.4986,
+    "eval_samples": 35,
+    "eval_samples_per_second": 5.386,
+    "eval_steps_per_second": 0.308,
+    "perplexity": 65.60118435636834,
+    "train_loss": 3.7094925158997714,
+    "train_runtime": 108243.5499,
+    "train_samples": 311198,
+    "train_samples_per_second": 2.875,
+    "train_steps_per_second": 0.045
+}

checkpoint-4000/config.json

@@ -0,0 +1,26 @@
+{
+  "_name_or_path": "/data/searchgpt/yq/GoGPT/outputs-pt-v1-7b-llama2/ckpt",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.29.1",
+  "use_cache": false,
+  "vocab_size": 68419
+}

checkpoint-4000/global_step4000/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:814e8129fe526c42fdbd6fe4ff2c9b21f0c161f6982e3883c3e5fea1260a4cb1
+size 2607759360

checkpoint-4000/global_step4000/zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aabd69e33896e791496a8b91b6325d7e6dafe50483e2ed374bf68f6bbc025924
+size 840736473

checkpoint-4000/pytorch_model-00001-of-00002.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcb73d2c6cc7cec148b9ea0157637718f3414f3a18d493daf4255e47befedae9
+size 10531361877

checkpoint-4000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7f1e51f8b260af2143aa0e342ab171171f7d983fbaaf95e9400d872b4d8c542
+size 21687

checkpoint-4000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:057591ad3bc33c9f75b1e6556c3c29b4328e0785d67a420f2a5a8bb25e58812e
+size 21687

checkpoint-4000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b595d89f2fe53ca5dccc290b880b858a60eb07a5f694175f05519f2257fafbf
+size 21687

checkpoint-4000/rng_state_4.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8dae996da909b80faa4e9fef9f945126f53fa3c7c6da06367a3f71ae24b0f138
+size 21687

checkpoint-4000/rng_state_5.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8de190b28a07218f4d05a1c7ccfccc4e53d6033af244fe3bd9fe3e20275d59de
+size 21687

checkpoint-4000/rng_state_6.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4cccb61a1758c3ec09da0ddc7ef0957d32362d5baa800f340839c1fa72376e2
+size 21687

checkpoint-4000/rng_state_7.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed085eb5cf063033df209f9c7714e3c87ee78a57b9ee629a7258a15bd8cde7ac
+size 21687

checkpoint-4000/trainer_state.json

@@ -0,0 +1,3142 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8226221079691517,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0,
+      "loss": 11.981,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9970117109141705e-05,
+      "loss": 12.0789,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.331893267009234e-05,
+      "loss": 10.7133,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011407670594843083,
+      "loss": 8.7339,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012721122651399258,
+      "loss": 8.301,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013684136855727938,
+      "loss": 8.1964,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "eval_accuracy": 0.10310832025117739,
+      "eval_loss": 8.203125,
+      "eval_runtime": 6.5764,
+      "eval_samples_per_second": 5.322,
+      "eval_steps_per_second": 0.304,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014444862339428802,
+      "loss": 8.0553,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015073705430110066,
+      "loss": 7.9436,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015609707636042195,
+      "loss": 7.8368,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016076788727202945,
+      "loss": 7.7333,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016490670495758757,
+      "loss": 7.6139,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "eval_accuracy": 0.12459794174079888,
+      "eval_loss": 7.81640625,
+      "eval_runtime": 6.6059,
+      "eval_samples_per_second": 5.298,
+      "eval_steps_per_second": 0.303,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001686224178807056,
+      "loss": 7.4892,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000171993565594773,
+      "loss": 7.3256,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017507866443784335,
+      "loss": 7.1827,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001779224840062419,
+      "loss": 6.9698,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018056004207494319,
+      "loss": 6.8162,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "eval_accuracy": 0.1567556253270539,
+      "eval_loss": 7.08203125,
+      "eval_runtime": 6.5908,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018301924610008189,
+      "loss": 6.6293,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018532269677939782,
+      "loss": 6.4114,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018748895370481112,
+      "loss": 6.2911,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018953344483335556,
+      "loss": 6.1047,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019146913367833817,
+      "loss": 5.9957,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "eval_accuracy": 0.19811616954474098,
+      "eval_loss": 6.4296875,
+      "eval_runtime": 6.5897,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.304,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019330701776944063,
+      "loss": 5.8281,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019505650713185044,
+      "loss": 5.6927,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019672571585424665,
+      "loss": 5.5564,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019832168964685297,
+      "loss": 5.3813,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998505855457085,
+      "loss": 5.2496,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "eval_accuracy": 0.24055817198674342,
+      "eval_loss": 5.8203125,
+      "eval_runtime": 6.6013,
+      "eval_samples_per_second": 5.302,
+      "eval_steps_per_second": 0.303,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996535296665223,
+      "loss": 5.13,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001992204417496752,
+      "loss": 5.0354,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019878735383282807,
+      "loss": 4.9021,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019835426591598097,
+      "loss": 4.8181,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019792117799913384,
+      "loss": 4.6993,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "eval_accuracy": 0.2641758241758242,
+      "eval_loss": 5.41796875,
+      "eval_runtime": 6.6009,
+      "eval_samples_per_second": 5.302,
+      "eval_steps_per_second": 0.303,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001974880900822867,
+      "loss": 4.6761,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019705500216543958,
+      "loss": 4.5908,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019662191424859245,
+      "loss": 4.5301,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019618882633174535,
+      "loss": 4.4729,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019575573841489822,
+      "loss": 4.3928,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "eval_accuracy": 0.27927437641723357,
+      "eval_loss": 5.14453125,
+      "eval_runtime": 6.6101,
+      "eval_samples_per_second": 5.295,
+      "eval_steps_per_second": 0.303,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019532265049805112,
+      "loss": 4.4083,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000194889562581204,
+      "loss": 4.336,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001944564746643569,
+      "loss": 4.2714,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019402338674750976,
+      "loss": 4.2124,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019359029883066263,
+      "loss": 4.2395,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "eval_accuracy": 0.2894470608756323,
+      "eval_loss": 4.96875,
+      "eval_runtime": 6.5945,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001931572109138155,
+      "loss": 4.1867,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001927241229969684,
+      "loss": 4.1687,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019229103508012127,
+      "loss": 4.1027,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019185794716327414,
+      "loss": 4.1233,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019142485924642704,
+      "loss": 4.0781,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "eval_accuracy": 0.29638932496075354,
+      "eval_loss": 4.8515625,
+      "eval_runtime": 6.5996,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001909917713295799,
+      "loss": 4.0855,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001905586834127328,
+      "loss": 4.0859,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019012559549588568,
+      "loss": 4.0124,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00018969250757903855,
+      "loss": 4.0151,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00018925941966219142,
+      "loss": 4.0409,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "eval_accuracy": 0.30181057038199893,
+      "eval_loss": 4.76953125,
+      "eval_runtime": 6.6117,
+      "eval_samples_per_second": 5.294,
+      "eval_steps_per_second": 0.302,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00018882633174534431,
+      "loss": 3.9912,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018839324382849718,
+      "loss": 3.9383,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018796015591165008,
+      "loss": 3.9764,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018752706799480295,
+      "loss": 3.9809,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018709398007795585,
+      "loss": 3.9178,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "eval_accuracy": 0.30575963718820864,
+      "eval_loss": 4.703125,
+      "eval_runtime": 6.6265,
+      "eval_samples_per_second": 5.282,
+      "eval_steps_per_second": 0.302,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018666089216110872,
+      "loss": 3.9073,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001862278042442616,
+      "loss": 3.9459,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018579471632741446,
+      "loss": 3.9535,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018536162841056733,
+      "loss": 3.8982,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018492854049372023,
+      "loss": 3.834,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "eval_accuracy": 0.3082574568288854,
+      "eval_loss": 4.65625,
+      "eval_runtime": 6.6135,
+      "eval_samples_per_second": 5.292,
+      "eval_steps_per_second": 0.302,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001844954525768731,
+      "loss": 3.8759,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000184062364660026,
+      "loss": 3.8835,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00018362927674317887,
+      "loss": 3.9003,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00018319618882633177,
+      "loss": 3.8538,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00018276310090948464,
+      "loss": 3.8316,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "eval_accuracy": 0.30981336124193265,
+      "eval_loss": 4.625,
+      "eval_runtime": 6.6039,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018233001299263754,
+      "loss": 3.9085,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018189692507579038,
+      "loss": 3.7825,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018146383715894328,
+      "loss": 3.824,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018103074924209615,
+      "loss": 3.8457,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018059766132524902,
+      "loss": 3.8197,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "eval_accuracy": 0.3116902145473574,
+      "eval_loss": 4.59765625,
+      "eval_runtime": 6.6155,
+      "eval_samples_per_second": 5.291,
+      "eval_steps_per_second": 0.302,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018016457340840192,
+      "loss": 3.79,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017973148549155479,
+      "loss": 3.7907,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017929839757470768,
+      "loss": 3.7797,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017886530965786055,
+      "loss": 3.7533,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017843222174101345,
+      "loss": 3.7464,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "eval_accuracy": 0.31469038897610324,
+      "eval_loss": 4.5625,
+      "eval_runtime": 6.5988,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017799913382416632,
+      "loss": 3.7347,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001775660459073192,
+      "loss": 3.7917,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017713295799047206,
+      "loss": 3.8106,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017669987007362496,
+      "loss": 3.7289,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017626678215677783,
+      "loss": 3.767,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "eval_accuracy": 0.3163718820861678,
+      "eval_loss": 4.5390625,
+      "eval_runtime": 6.604,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017583369423993073,
+      "loss": 3.7362,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001754006063230836,
+      "loss": 3.7474,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017496751840623647,
+      "loss": 3.7485,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017453443048938937,
+      "loss": 3.7341,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017410134257254224,
+      "loss": 3.7511,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "eval_accuracy": 0.3172649572649573,
+      "eval_loss": 4.515625,
+      "eval_runtime": 6.6093,
+      "eval_samples_per_second": 5.296,
+      "eval_steps_per_second": 0.303,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001736682546556951,
+      "loss": 3.6862,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017323516673884798,
+      "loss": 3.6411,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017280207882200088,
+      "loss": 3.7181,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017236899090515375,
+      "loss": 3.6471,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017193590298830665,
+      "loss": 3.7166,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "eval_accuracy": 0.3188278388278388,
+      "eval_loss": 4.4921875,
+      "eval_runtime": 6.6113,
+      "eval_samples_per_second": 5.294,
+      "eval_steps_per_second": 0.303,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017150281507145952,
+      "loss": 3.6869,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017106972715461241,
+      "loss": 3.6728,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017063663923776528,
+      "loss": 3.7705,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017020355132091815,
+      "loss": 3.6728,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00016977046340407103,
+      "loss": 3.6908,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "eval_accuracy": 0.3190441304727019,
+      "eval_loss": 4.48046875,
+      "eval_runtime": 6.5961,
+      "eval_samples_per_second": 5.306,
+      "eval_steps_per_second": 0.303,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001693373754872239,
+      "loss": 3.5988,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001689042875703768,
+      "loss": 3.6713,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00016847119965352966,
+      "loss": 3.7165,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00016803811173668256,
+      "loss": 3.7098,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016760502381983543,
+      "loss": 3.617,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "eval_accuracy": 0.3208442351299494,
+      "eval_loss": 4.46484375,
+      "eval_runtime": 6.5949,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016717193590298833,
+      "loss": 3.6629,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001667388479861412,
+      "loss": 3.684,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016630576006929407,
+      "loss": 3.6877,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016587267215244694,
+      "loss": 3.6274,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00016543958423559984,
+      "loss": 3.6734,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "eval_accuracy": 0.3213535670678528,
+      "eval_loss": 4.453125,
+      "eval_runtime": 6.5974,
+      "eval_samples_per_second": 5.305,
+      "eval_steps_per_second": 0.303,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001650064963187527,
+      "loss": 3.6895,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001645734084019056,
+      "loss": 3.6662,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00016414032048505848,
+      "loss": 3.6152,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00016370723256821135,
+      "loss": 3.6531,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016327414465136425,
+      "loss": 3.6916,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "eval_accuracy": 0.32398395255538115,
+      "eval_loss": 4.43359375,
+      "eval_runtime": 6.5897,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.304,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016284105673451712,
+      "loss": 3.6961,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016240796881767,
+      "loss": 3.5811,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016197488090082286,
+      "loss": 3.6161,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016154179298397576,
+      "loss": 3.6305,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016110870506712863,
+      "loss": 3.629,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "eval_accuracy": 0.32400488400488403,
+      "eval_loss": 4.421875,
+      "eval_runtime": 6.5985,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016067561715028152,
+      "loss": 3.6467,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001602425292334344,
+      "loss": 3.6573,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001598094413165873,
+      "loss": 3.6372,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00015937635339974016,
+      "loss": 3.6369,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00015894326548289303,
+      "loss": 3.6001,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "eval_accuracy": 0.3247165532879819,
+      "eval_loss": 4.4140625,
+      "eval_runtime": 6.602,
+      "eval_samples_per_second": 5.301,
+      "eval_steps_per_second": 0.303,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001585101775660459,
+      "loss": 3.5843,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00015807708964919877,
+      "loss": 3.6407,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00015764400173235167,
+      "loss": 3.6413,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00015721091381550454,
+      "loss": 3.5963,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00015677782589865744,
+      "loss": 3.6053,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "eval_accuracy": 0.32607709750566893,
+      "eval_loss": 4.40234375,
+      "eval_runtime": 6.5982,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001563447379818103,
+      "loss": 3.6049,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001559116500649632,
+      "loss": 3.6112,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00015547856214811608,
+      "loss": 3.5872,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015504547423126895,
+      "loss": 3.6328,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015461238631442182,
+      "loss": 3.5803,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "eval_accuracy": 0.32540729112157685,
+      "eval_loss": 4.390625,
+      "eval_runtime": 6.5995,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015417929839757472,
+      "loss": 3.6153,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001537462104807276,
+      "loss": 3.5493,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015331312256388049,
+      "loss": 3.5854,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015288003464703336,
+      "loss": 3.6029,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015244694673018623,
+      "loss": 3.5886,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "eval_accuracy": 0.32738880167451595,
+      "eval_loss": 4.37890625,
+      "eval_runtime": 6.5931,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015201385881333913,
+      "loss": 3.5496,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000151580770896492,
+      "loss": 3.557,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015114768297964487,
+      "loss": 3.5647,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015071459506279774,
+      "loss": 3.5912,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015028150714595063,
+      "loss": 3.5033,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "eval_accuracy": 0.3287981859410431,
+      "eval_loss": 4.3671875,
+      "eval_runtime": 6.593,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001498484192291035,
+      "loss": 3.5814,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001494153313122564,
+      "loss": 3.5834,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00014898224339540927,
+      "loss": 3.5661,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014854915547856217,
+      "loss": 3.5844,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014811606756171504,
+      "loss": 3.58,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "eval_accuracy": 0.3283656026513169,
+      "eval_loss": 4.36328125,
+      "eval_runtime": 6.6004,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001476829796448679,
+      "loss": 3.5874,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014724989172802078,
+      "loss": 3.5736,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014681680381117365,
+      "loss": 3.5659,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014638371589432655,
+      "loss": 3.5632,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014595062797747942,
+      "loss": 3.4966,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "eval_accuracy": 0.32832373975231116,
+      "eval_loss": 4.3515625,
+      "eval_runtime": 6.592,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014551754006063232,
+      "loss": 3.5236,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001450844521437852,
+      "loss": 3.5277,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001446513642269381,
+      "loss": 3.5237,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014421827631009096,
+      "loss": 3.5719,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014378518839324383,
+      "loss": 3.5411,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "eval_accuracy": 0.32884004884004886,
+      "eval_loss": 4.3515625,
+      "eval_runtime": 6.6086,
+      "eval_samples_per_second": 5.296,
+      "eval_steps_per_second": 0.303,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001433521004763967,
+      "loss": 3.5287,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001429190125595496,
+      "loss": 3.5965,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014248592464270247,
+      "loss": 3.5435,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014205283672585536,
+      "loss": 3.5536,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014161974880900824,
+      "loss": 3.527,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "eval_accuracy": 0.33027036455607883,
+      "eval_loss": 4.33984375,
+      "eval_runtime": 6.5917,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001411866608921611,
+      "loss": 3.5765,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000140753572975314,
+      "loss": 3.5882,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014032048505846687,
+      "loss": 3.5135,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013988739714161974,
+      "loss": 3.4924,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013945430922477262,
+      "loss": 3.6018,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "eval_accuracy": 0.3299284842141985,
+      "eval_loss": 4.33203125,
+      "eval_runtime": 6.5992,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001390212213079255,
+      "loss": 3.5646,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013858813339107838,
+      "loss": 3.5164,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013815504547423128,
+      "loss": 3.5433,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013772195755738415,
+      "loss": 3.4929,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013728886964053705,
+      "loss": 3.4802,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "eval_accuracy": 0.3301238444095587,
+      "eval_loss": 4.32421875,
+      "eval_runtime": 6.5899,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.303,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013685578172368992,
+      "loss": 3.5211,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013642269380684282,
+      "loss": 3.5424,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013598960588999566,
+      "loss": 3.468,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013555651797314853,
+      "loss": 3.5342,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013512343005630143,
+      "loss": 3.4375,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "eval_accuracy": 0.330584336298622,
+      "eval_loss": 4.32421875,
+      "eval_runtime": 6.5928,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001346903421394543,
+      "loss": 3.5239,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001342572542226072,
+      "loss": 3.5176,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013382416630576007,
+      "loss": 3.4992,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013339107838891297,
+      "loss": 3.457,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013295799047206584,
+      "loss": 3.4873,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "eval_accuracy": 0.33109366823652536,
+      "eval_loss": 4.3203125,
+      "eval_runtime": 6.6125,
+      "eval_samples_per_second": 5.293,
+      "eval_steps_per_second": 0.302,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013252490255521873,
+      "loss": 3.5021,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001320918146383716,
+      "loss": 3.4871,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013165872672152448,
+      "loss": 3.47,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013122563880467735,
+      "loss": 3.4462,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013079255088783024,
+      "loss": 3.435,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "eval_accuracy": 0.3309401709401709,
+      "eval_loss": 4.3125,
+      "eval_runtime": 6.5916,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013035946297098311,
+      "loss": 3.4994,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00012992637505413598,
+      "loss": 3.5665,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012949328713728888,
+      "loss": 3.5552,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012906019922044175,
+      "loss": 3.5092,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012862711130359465,
+      "loss": 3.4335,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "eval_accuracy": 0.3317634746206175,
+      "eval_loss": 4.3046875,
+      "eval_runtime": 6.6183,
+      "eval_samples_per_second": 5.288,
+      "eval_steps_per_second": 0.302,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012819402338674752,
+      "loss": 3.5154,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001277609354699004,
+      "loss": 3.5207,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012732784755305326,
+      "loss": 3.4869,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012689475963620616,
+      "loss": 3.4773,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012646167171935903,
+      "loss": 3.4595,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "eval_accuracy": 0.3325239839525554,
+      "eval_loss": 4.296875,
+      "eval_runtime": 6.5968,
+      "eval_samples_per_second": 5.306,
+      "eval_steps_per_second": 0.303,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012602858380251193,
+      "loss": 3.5035,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001255954958856648,
+      "loss": 3.4692,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001251624079688177,
+      "loss": 3.4712,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012472932005197057,
+      "loss": 3.4558,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012429623213512344,
+      "loss": 3.4937,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "eval_accuracy": 0.3319448805163091,
+      "eval_loss": 4.29296875,
+      "eval_runtime": 6.5908,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001238631442182763,
+      "loss": 3.5093,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012343005630142918,
+      "loss": 3.4941,
+      "step": 2020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012299696838458208,
+      "loss": 3.4706,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012256388046773495,
+      "loss": 3.5106,
+      "step": 2040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012213079255088784,
+      "loss": 3.4959,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "eval_accuracy": 0.3324123495552067,
+      "eval_loss": 4.28515625,
+      "eval_runtime": 6.6003,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012169770463404072,
+      "loss": 3.5218,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001212646167171936,
+      "loss": 3.4813,
+      "step": 2070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012083152880034647,
+      "loss": 3.5107,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012039844088349937,
+      "loss": 3.4568,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011996535296665224,
+      "loss": 3.4987,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "eval_accuracy": 0.3331658817373103,
+      "eval_loss": 4.28515625,
+      "eval_runtime": 6.6107,
+      "eval_samples_per_second": 5.294,
+      "eval_steps_per_second": 0.303,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011957557384148984,
+      "loss": 3.458,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001191424859246427,
+      "loss": 3.4656,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011870939800779559,
+      "loss": 3.4505,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011827631009094846,
+      "loss": 3.4182,
+      "step": 2140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011784322217410136,
+      "loss": 3.4001,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "eval_accuracy": 0.3336403279260422,
+      "eval_loss": 4.28515625,
+      "eval_runtime": 6.5938,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011741013425725423,
+      "loss": 3.4289,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011697704634040711,
+      "loss": 3.4228,
+      "step": 2170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011654395842355998,
+      "loss": 3.4066,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011611087050671288,
+      "loss": 3.4823,
+      "step": 2190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011567778258986575,
+      "loss": 3.4497,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "eval_accuracy": 0.3340380254665969,
+      "eval_loss": 4.28125,
+      "eval_runtime": 6.606,
+      "eval_samples_per_second": 5.298,
+      "eval_steps_per_second": 0.303,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011524469467301864,
+      "loss": 3.4737,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001148116067561715,
+      "loss": 3.4564,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011437851883932438,
+      "loss": 3.4865,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011394543092247727,
+      "loss": 3.4496,
+      "step": 2240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011351234300563015,
+      "loss": 3.4068,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "eval_accuracy": 0.33329147043432755,
+      "eval_loss": 4.27734375,
+      "eval_runtime": 6.6133,
+      "eval_samples_per_second": 5.292,
+      "eval_steps_per_second": 0.302,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011307925508878303,
+      "loss": 3.4337,
+      "step": 2260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001126461671719359,
+      "loss": 3.4338,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001122130792550888,
+      "loss": 3.4241,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011177999133824167,
+      "loss": 3.497,
+      "step": 2290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011134690342139455,
+      "loss": 3.4634,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "eval_accuracy": 0.3339822082679226,
+      "eval_loss": 4.2734375,
+      "eval_runtime": 6.5925,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011091381550454742,
+      "loss": 3.4245,
+      "step": 2310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011048072758770032,
+      "loss": 3.4465,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011004763967085319,
+      "loss": 3.498,
+      "step": 2330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010961455175400607,
+      "loss": 3.3637,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010918146383715895,
+      "loss": 3.4324,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "eval_accuracy": 0.333849642421071,
+      "eval_loss": 4.27734375,
+      "eval_runtime": 6.5996,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010874837592031182,
+      "loss": 3.3816,
+      "step": 2360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010831528800346471,
+      "loss": 3.4245,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010788220008661758,
+      "loss": 3.4567,
+      "step": 2380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010744911216977047,
+      "loss": 3.4122,
+      "step": 2390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010701602425292334,
+      "loss": 3.4039,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "eval_accuracy": 0.3344147915576487,
+      "eval_loss": 4.265625,
+      "eval_runtime": 6.5992,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010658293633607624,
+      "loss": 3.4124,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010614984841922911,
+      "loss": 3.4309,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010571676050238199,
+      "loss": 3.4464,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010528367258553486,
+      "loss": 3.4136,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010485058466868776,
+      "loss": 3.4502,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "eval_accuracy": 0.33453340310483165,
+      "eval_loss": 4.265625,
+      "eval_runtime": 6.7512,
+      "eval_samples_per_second": 5.184,
+      "eval_steps_per_second": 0.296,
+      "step": 2450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010441749675184063,
+      "loss": 3.4062,
+      "step": 2460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010398440883499351,
+      "loss": 3.3951,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010355132091814638,
+      "loss": 3.4322,
+      "step": 2480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010311823300129926,
+      "loss": 3.4107,
+      "step": 2490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010268514508445215,
+      "loss": 3.4104,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "eval_accuracy": 0.33495900924472355,
+      "eval_loss": 4.2578125,
+      "eval_runtime": 6.5889,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 2500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010225205716760502,
+      "loss": 3.4181,
+      "step": 2510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010181896925075791,
+      "loss": 3.3979,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010138588133391078,
+      "loss": 3.3911,
+      "step": 2530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010095279341706368,
+      "loss": 3.4303,
+      "step": 2540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010051970550021655,
+      "loss": 3.5251,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "eval_accuracy": 0.33599860457003317,
+      "eval_loss": 4.2578125,
+      "eval_runtime": 6.5946,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 2550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010008661758336943,
+      "loss": 3.4337,
+      "step": 2560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96535296665223e-05,
+      "loss": 3.3906,
+      "step": 2570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.922044174967519e-05,
+      "loss": 3.4714,
+      "step": 2580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.878735383282807e-05,
+      "loss": 3.4061,
+      "step": 2590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.835426591598095e-05,
+      "loss": 3.4176,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "eval_accuracy": 0.3363404849119135,
+      "eval_loss": 4.25,
+      "eval_runtime": 6.594,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 2600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.792117799913382e-05,
+      "loss": 3.37,
+      "step": 2610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.748809008228671e-05,
+      "loss": 3.3931,
+      "step": 2620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.705500216543959e-05,
+      "loss": 3.4052,
+      "step": 2630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.662191424859248e-05,
+      "loss": 3.4269,
+      "step": 2640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.618882633174535e-05,
+      "loss": 3.3795,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "eval_accuracy": 0.3354404325832897,
+      "eval_loss": 4.25,
+      "eval_runtime": 6.6039,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 2650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.575573841489823e-05,
+      "loss": 3.4064,
+      "step": 2660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.532265049805112e-05,
+      "loss": 3.4207,
+      "step": 2670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.488956258120399e-05,
+      "loss": 3.4353,
+      "step": 2680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.445647466435687e-05,
+      "loss": 3.4497,
+      "step": 2690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.402338674750974e-05,
+      "loss": 3.3656,
+      "step": 2700
+    },
+    {
+      "epoch": 0.56,
+      "eval_accuracy": 0.33636839351125064,
+      "eval_loss": 4.25,
+      "eval_runtime": 6.5948,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 2700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.359029883066262e-05,
+      "loss": 3.3736,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.315721091381551e-05,
+      "loss": 3.4236,
+      "step": 2720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.272412299696839e-05,
+      "loss": 3.4234,
+      "step": 2730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.229103508012126e-05,
+      "loss": 3.3849,
+      "step": 2740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.185794716327415e-05,
+      "loss": 3.3938,
+      "step": 2750
+    },
+    {
+      "epoch": 0.57,
+      "eval_accuracy": 0.33627769056340484,
+      "eval_loss": 4.24609375,
+      "eval_runtime": 6.5953,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 2750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.142485924642703e-05,
+      "loss": 3.375,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.099177132957992e-05,
+      "loss": 3.4365,
+      "step": 2770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.055868341273279e-05,
+      "loss": 3.4068,
+      "step": 2780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.012559549588567e-05,
+      "loss": 3.4333,
+      "step": 2790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.969250757903855e-05,
+      "loss": 3.3757,
+      "step": 2800
+    },
+    {
+      "epoch": 0.58,
+      "eval_accuracy": 0.33639630211058785,
+      "eval_loss": 4.24609375,
+      "eval_runtime": 6.5987,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 2800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.925941966219143e-05,
+      "loss": 3.4111,
+      "step": 2810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.882633174534431e-05,
+      "loss": 3.4143,
+      "step": 2820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.839324382849718e-05,
+      "loss": 3.4307,
+      "step": 2830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.796015591165006e-05,
+      "loss": 3.3366,
+      "step": 2840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.752706799480295e-05,
+      "loss": 3.407,
+      "step": 2850
+    },
+    {
+      "epoch": 0.59,
+      "eval_accuracy": 0.33731728588871446,
+      "eval_loss": 4.234375,
+      "eval_runtime": 6.6038,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 2850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.709398007795583e-05,
+      "loss": 3.3506,
+      "step": 2860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.66608921611087e-05,
+      "loss": 3.429,
+      "step": 2870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.622780424426159e-05,
+      "loss": 3.338,
+      "step": 2880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.579471632741447e-05,
+      "loss": 3.4252,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.536162841056736e-05,
+      "loss": 3.3986,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6,
+      "eval_accuracy": 0.33657770800627945,
+      "eval_loss": 4.23828125,
+      "eval_runtime": 6.5874,
+      "eval_samples_per_second": 5.313,
+      "eval_steps_per_second": 0.304,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.492854049372023e-05,
+      "loss": 3.4023,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.449545257687311e-05,
+      "loss": 3.3919,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.4062364660026e-05,
+      "loss": 3.3691,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.362927674317888e-05,
+      "loss": 3.4017,
+      "step": 2940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.319618882633175e-05,
+      "loss": 3.4311,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "eval_accuracy": 0.3370940170940171,
+      "eval_loss": 4.234375,
+      "eval_runtime": 6.5922,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.276310090948462e-05,
+      "loss": 3.4224,
+      "step": 2960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.23300129926375e-05,
+      "loss": 3.3781,
+      "step": 2970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.189692507579039e-05,
+      "loss": 3.383,
+      "step": 2980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.146383715894327e-05,
+      "loss": 3.4332,
+      "step": 2990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.103074924209616e-05,
+      "loss": 3.3716,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "eval_accuracy": 0.33713587999302286,
+      "eval_loss": 4.234375,
+      "eval_runtime": 6.5875,
+      "eval_samples_per_second": 5.313,
+      "eval_steps_per_second": 0.304,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.059766132524903e-05,
+      "loss": 3.4123,
+      "step": 3010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.016457340840191e-05,
+      "loss": 3.4181,
+      "step": 3020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.97314854915548e-05,
+      "loss": 3.3851,
+      "step": 3030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.929839757470768e-05,
+      "loss": 3.4224,
+      "step": 3040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.886530965786055e-05,
+      "loss": 3.3831,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "eval_accuracy": 0.33774986917844063,
+      "eval_loss": 4.23046875,
+      "eval_runtime": 6.5895,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.304,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.843222174101343e-05,
+      "loss": 3.3965,
+      "step": 3060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.799913382416632e-05,
+      "loss": 3.4016,
+      "step": 3070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.756604590731919e-05,
+      "loss": 3.4047,
+      "step": 3080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.713295799047207e-05,
+      "loss": 3.3618,
+      "step": 3090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.669987007362494e-05,
+      "loss": 3.375,
+      "step": 3100
+    },
+    {
+      "epoch": 0.64,
+      "eval_accuracy": 0.3377638234781092,
+      "eval_loss": 4.23046875,
+      "eval_runtime": 6.5887,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 3100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.626678215677783e-05,
+      "loss": 3.3899,
+      "step": 3110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.583369423993071e-05,
+      "loss": 3.3723,
+      "step": 3120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.54006063230836e-05,
+      "loss": 3.381,
+      "step": 3130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.496751840623647e-05,
+      "loss": 3.3558,
+      "step": 3140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.453443048938935e-05,
+      "loss": 3.3677,
+      "step": 3150
+    },
+    {
+      "epoch": 0.65,
+      "eval_accuracy": 0.33846851561137276,
+      "eval_loss": 4.21875,
+      "eval_runtime": 6.5962,
+      "eval_samples_per_second": 5.306,
+      "eval_steps_per_second": 0.303,
+      "step": 3150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.410134257254223e-05,
+      "loss": 3.3414,
+      "step": 3160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.366825465569512e-05,
+      "loss": 3.4144,
+      "step": 3170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.323516673884799e-05,
+      "loss": 3.3503,
+      "step": 3180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.280207882200087e-05,
+      "loss": 3.3716,
+      "step": 3190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.236899090515376e-05,
+      "loss": 3.3968,
+      "step": 3200
+    },
+    {
+      "epoch": 0.66,
+      "eval_accuracy": 0.3386220129077272,
+      "eval_loss": 4.22265625,
+      "eval_runtime": 6.5929,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 3200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.193590298830663e-05,
+      "loss": 3.3485,
+      "step": 3210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.150281507145951e-05,
+      "loss": 3.4172,
+      "step": 3220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.106972715461238e-05,
+      "loss": 3.3727,
+      "step": 3230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.063663923776527e-05,
+      "loss": 3.3616,
+      "step": 3240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.020355132091815e-05,
+      "loss": 3.4069,
+      "step": 3250
+    },
+    {
+      "epoch": 0.67,
+      "eval_accuracy": 0.3380429094714809,
+      "eval_loss": 4.21875,
+      "eval_runtime": 6.5865,
+      "eval_samples_per_second": 5.314,
+      "eval_steps_per_second": 0.304,
+      "step": 3250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.977046340407103e-05,
+      "loss": 3.3583,
+      "step": 3260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.93373754872239e-05,
+      "loss": 3.3753,
+      "step": 3270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.890428757037679e-05,
+      "loss": 3.3443,
+      "step": 3280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.847119965352967e-05,
+      "loss": 3.3682,
+      "step": 3290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.803811173668256e-05,
+      "loss": 3.4192,
+      "step": 3300
+    },
+    {
+      "epoch": 0.68,
+      "eval_accuracy": 0.33877551020408164,
+      "eval_loss": 4.21484375,
+      "eval_runtime": 6.6033,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 3300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.760502381983543e-05,
+      "loss": 3.3657,
+      "step": 3310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.717193590298831e-05,
+      "loss": 3.3773,
+      "step": 3320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.67388479861412e-05,
+      "loss": 3.3604,
+      "step": 3330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.630576006929407e-05,
+      "loss": 3.404,
+      "step": 3340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.587267215244695e-05,
+      "loss": 3.3881,
+      "step": 3350
+    },
+    {
+      "epoch": 0.69,
+      "eval_accuracy": 0.33830804116518404,
+      "eval_loss": 4.21484375,
+      "eval_runtime": 6.6138,
+      "eval_samples_per_second": 5.292,
+      "eval_steps_per_second": 0.302,
+      "step": 3350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.543958423559982e-05,
+      "loss": 3.3855,
+      "step": 3360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.50064963187527e-05,
+      "loss": 3.314,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.457340840190559e-05,
+      "loss": 3.4034,
+      "step": 3380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.414032048505847e-05,
+      "loss": 3.3969,
+      "step": 3390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.370723256821134e-05,
+      "loss": 3.3858,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7,
+      "eval_accuracy": 0.33836385836385835,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.6099,
+      "eval_samples_per_second": 5.295,
+      "eval_steps_per_second": 0.303,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.327414465136423e-05,
+      "loss": 3.388,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.284105673451711e-05,
+      "loss": 3.3832,
+      "step": 3420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.240796881767e-05,
+      "loss": 3.3098,
+      "step": 3430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.197488090082287e-05,
+      "loss": 3.3794,
+      "step": 3440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.154179298397575e-05,
+      "loss": 3.3999,
+      "step": 3450
+    },
+    {
+      "epoch": 0.71,
+      "eval_accuracy": 0.33883132740275596,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.6039,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 3450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.110870506712864e-05,
+      "loss": 3.3278,
+      "step": 3460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0675617150281506e-05,
+      "loss": 3.3645,
+      "step": 3470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.024252923343438e-05,
+      "loss": 3.3563,
+      "step": 3480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.985275010827198e-05,
+      "loss": 3.358,
+      "step": 3490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9419662191424864e-05,
+      "loss": 3.3907,
+      "step": 3500
+    },
+    {
+      "epoch": 0.72,
+      "eval_accuracy": 0.33894296180010464,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.6054,
+      "eval_samples_per_second": 5.299,
+      "eval_steps_per_second": 0.303,
+      "step": 3500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.898657427457774e-05,
+      "loss": 3.4023,
+      "step": 3510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8553486357730626e-05,
+      "loss": 3.3519,
+      "step": 3520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.81203984408835e-05,
+      "loss": 3.3514,
+      "step": 3530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.768731052403639e-05,
+      "loss": 3.3151,
+      "step": 3540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.7254222607189265e-05,
+      "loss": 3.3929,
+      "step": 3550
+    },
+    {
+      "epoch": 0.73,
+      "eval_accuracy": 0.33893598465027036,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.5935,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 3550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6821134690342135e-05,
+      "loss": 3.3983,
+      "step": 3560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.638804677349502e-05,
+      "loss": 3.3281,
+      "step": 3570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.59549588566479e-05,
+      "loss": 3.3654,
+      "step": 3580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.552187093980078e-05,
+      "loss": 3.379,
+      "step": 3590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.508878302295366e-05,
+      "loss": 3.3738,
+      "step": 3600
+    },
+    {
+      "epoch": 0.74,
+      "eval_accuracy": 0.33964067678353393,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.6063,
+      "eval_samples_per_second": 5.298,
+      "eval_steps_per_second": 0.303,
+      "step": 3600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.465569510610654e-05,
+      "loss": 3.3727,
+      "step": 3610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.426591598094414e-05,
+      "loss": 3.3401,
+      "step": 3620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.3832828064097017e-05,
+      "loss": 3.3583,
+      "step": 3630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.33997401472499e-05,
+      "loss": 3.3868,
+      "step": 3640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.296665223040277e-05,
+      "loss": 3.3839,
+      "step": 3650
+    },
+    {
+      "epoch": 0.75,
+      "eval_accuracy": 0.3392918192918193,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.5952,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 3650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.253356431355565e-05,
+      "loss": 3.3713,
+      "step": 3660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.210047639670853e-05,
+      "loss": 3.3649,
+      "step": 3670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.166738847986141e-05,
+      "loss": 3.3836,
+      "step": 3680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.1234300563014294e-05,
+      "loss": 3.3227,
+      "step": 3690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.080121264616717e-05,
+      "loss": 3.3854,
+      "step": 3700
+    },
+    {
+      "epoch": 0.76,
+      "eval_accuracy": 0.3399546485260771,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.6101,
+      "eval_samples_per_second": 5.295,
+      "eval_steps_per_second": 0.303,
+      "step": 3700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0368124729320056e-05,
+      "loss": 3.3536,
+      "step": 3710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993503681247293e-05,
+      "loss": 3.3943,
+      "step": 3720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.950194889562582e-05,
+      "loss": 3.3705,
+      "step": 3730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9068860978778694e-05,
+      "loss": 3.3525,
+      "step": 3740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.863577306193157e-05,
+      "loss": 3.297,
+      "step": 3750
+    },
+    {
+      "epoch": 0.77,
+      "eval_accuracy": 0.339236002093145,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.5942,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 3750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8202685145084456e-05,
+      "loss": 3.3305,
+      "step": 3760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.776959722823733e-05,
+      "loss": 3.3789,
+      "step": 3770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.733650931139022e-05,
+      "loss": 3.4127,
+      "step": 3780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6903421394543095e-05,
+      "loss": 3.3234,
+      "step": 3790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.647033347769598e-05,
+      "loss": 3.2951,
+      "step": 3800
+    },
+    {
+      "epoch": 0.78,
+      "eval_accuracy": 0.3394732251875109,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.5899,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.303,
+      "step": 3800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.603724556084885e-05,
+      "loss": 3.3538,
+      "step": 3810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5604157644001733e-05,
+      "loss": 3.4101,
+      "step": 3820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.517106972715461e-05,
+      "loss": 3.3232,
+      "step": 3830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4737981810307495e-05,
+      "loss": 3.3519,
+      "step": 3840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.430489389346037e-05,
+      "loss": 3.3587,
+      "step": 3850
+    },
+    {
+      "epoch": 0.79,
+      "eval_accuracy": 0.3401709401709402,
+      "eval_loss": 4.19921875,
+      "eval_runtime": 6.5915,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 3850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3871805976613256e-05,
+      "loss": 3.3519,
+      "step": 3860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.3438718059766134e-05,
+      "loss": 3.3635,
+      "step": 3870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.300563014291901e-05,
+      "loss": 3.4013,
+      "step": 3880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2572542226071895e-05,
+      "loss": 3.3224,
+      "step": 3890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.213945430922477e-05,
+      "loss": 3.3237,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8,
+      "eval_accuracy": 0.33935461364032793,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.5894,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.170636639237766e-05,
+      "loss": 3.3706,
+      "step": 3910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1273278475530534e-05,
+      "loss": 3.3024,
+      "step": 3920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.084019055868342e-05,
+      "loss": 3.3717,
+      "step": 3930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0407102641836295e-05,
+      "loss": 3.35,
+      "step": 3940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.997401472498917e-05,
+      "loss": 3.3136,
+      "step": 3950
+    },
+    {
+      "epoch": 0.81,
+      "eval_accuracy": 0.3393267050409908,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.6033,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 3950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.954092680814206e-05,
+      "loss": 3.3281,
+      "step": 3960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9107838891294934e-05,
+      "loss": 3.2916,
+      "step": 3970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.867475097444782e-05,
+      "loss": 3.3578,
+      "step": 3980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8241663057600696e-05,
+      "loss": 3.3789,
+      "step": 3990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.780857514075358e-05,
+      "loss": 3.3367,
+      "step": 4000
+    },
+    {
+      "epoch": 0.82,
+      "eval_accuracy": 0.33942438513867085,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.5852,
+      "eval_samples_per_second": 5.315,
+      "eval_steps_per_second": 0.304,
+      "step": 4000
+    }
+  ],
+  "max_steps": 4862,
+  "num_train_epochs": 1,
+  "total_flos": 4.250832212066304e+19,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-4000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8882a918bcb9a315dfcc77f1a3a879de69b8b68f463ac4b04ba7f0c085598f2d
+size 5115

checkpoint-4000/zero_to_fp32.py

@@ -0,0 +1,578 @@
+#!/usr/bin/env python
+
+# Copyright (c) Microsoft Corporation.
+# SPDX-License-Identifier: Apache-2.0
+
+# DeepSpeed Team
+
+# This script extracts fp32 consolidated weights from a zero 2 and 3 DeepSpeed checkpoints. It gets
+# copied into the top level checkpoint dir, so the user can easily do the conversion at any point in
+# the future. Once extracted, the weights don't require DeepSpeed and can be used in any
+# application.
+#
+# example: python zero_to_fp32.py . pytorch_model.bin
+
+import argparse
+import torch
+import glob
+import math
+import os
+import re
+from collections import OrderedDict
+from dataclasses import dataclass
+
+# while this script doesn't use deepspeed to recover data, since the checkpoints are pickled with
+# DeepSpeed data structures it has to be available in the current python environment.
+from deepspeed.utils import logger
+from deepspeed.checkpoint.constants import (DS_VERSION, OPTIMIZER_STATE_DICT, SINGLE_PARTITION_OF_FP32_GROUPS,
+                                            FP32_FLAT_GROUPS, ZERO_STAGE, PARTITION_COUNT, PARAM_SHAPES, BUFFER_NAMES,
+                                            FROZEN_PARAM_SHAPES, FROZEN_PARAM_FRAGMENTS)
+
+
+@dataclass
+class zero_model_state:
+    buffers: dict()
+    param_shapes: dict()
+    shared_params: list
+    ds_version: int
+    frozen_param_shapes: dict()
+    frozen_param_fragments: dict()
+
+
+debug = 0
+
+# load to cpu
+device = torch.device('cpu')
+
+
+def atoi(text):
+    return int(text) if text.isdigit() else text
+
+
+def natural_keys(text):
+    '''
+    alist.sort(key=natural_keys) sorts in human order
+    http://nedbatchelder.com/blog/200712/human_sorting.html
+    (See Toothy's implementation in the comments)
+    '''
+    return [atoi(c) for c in re.split(r'(\d+)', text)]
+
+
+def get_model_state_file(checkpoint_dir, zero_stage):
+    if not os.path.isdir(checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{checkpoint_dir}' doesn't exist")
+
+    # there should be only one file
+    if zero_stage == 2:
+        file = os.path.join(checkpoint_dir, "mp_rank_00_model_states.pt")
+    elif zero_stage == 3:
+        file = os.path.join(checkpoint_dir, "zero_pp_rank_0_mp_rank_00_model_states.pt")
+
+    if not os.path.exists(file):
+        raise FileNotFoundError(f"can't find model states file at '{file}'")
+
+    return file
+
+
+def get_checkpoint_files(checkpoint_dir, glob_pattern):
+    # XXX: need to test that this simple glob rule works for multi-node setup too
+    ckpt_files = sorted(glob.glob(os.path.join(checkpoint_dir, glob_pattern)), key=natural_keys)
+
+    if len(ckpt_files) == 0:
+        raise FileNotFoundError(f"can't find {glob_pattern} files in directory '{checkpoint_dir}'")
+
+    return ckpt_files
+
+
+def get_optim_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_optim_states.pt")
+
+
+def get_model_state_files(checkpoint_dir):
+    return get_checkpoint_files(checkpoint_dir, "*_model_states.pt")
+
+
+def parse_model_states(files):
+    zero_model_states = []
+    for file in files:
+        state_dict = torch.load(file, map_location=device)
+
+        if BUFFER_NAMES not in state_dict:
+            raise ValueError(f"{file} is not a model state checkpoint")
+        buffer_names = state_dict[BUFFER_NAMES]
+        if debug:
+            print("Found buffers:", buffer_names)
+
+        # recover just the buffers while restoring them to fp32 if they were saved in fp16
+        buffers = {k: v.float() for k, v in state_dict["module"].items() if k in buffer_names}
+        param_shapes = state_dict[PARAM_SHAPES]
+
+        # collect parameters that are included in param_shapes
+        param_names = []
+        for s in param_shapes:
+            for name in s.keys():
+                param_names.append(name)
+
+        # update with frozen parameters
+        frozen_param_shapes = state_dict.get(FROZEN_PARAM_SHAPES, None)
+        if frozen_param_shapes is not None:
+            if debug:
+                print(f"Found frozen_param_shapes: {frozen_param_shapes}")
+            param_names += list(frozen_param_shapes.keys())
+
+        # handle shared params
+        shared_params = [[k, v] for k, v in state_dict["shared_params"].items()]
+
+        ds_version = state_dict.get(DS_VERSION, None)
+
+        frozen_param_fragments = state_dict.get(FROZEN_PARAM_FRAGMENTS, None)
+
+        z_model_state = zero_model_state(buffers=buffers,
+                                         param_shapes=param_shapes,
+                                         shared_params=shared_params,
+                                         ds_version=ds_version,
+                                         frozen_param_shapes=frozen_param_shapes,
+                                         frozen_param_fragments=frozen_param_fragments)
+        zero_model_states.append(z_model_state)
+
+    return zero_model_states
+
+
+def parse_optim_states(files, ds_checkpoint_dir):
+
+    total_files = len(files)
+    state_dicts = []
+    for f in files:
+        state_dicts.append(torch.load(f, map_location=device))
+
+    if not ZERO_STAGE in state_dicts[0][OPTIMIZER_STATE_DICT]:
+        raise ValueError(f"{files[0]} is not a zero checkpoint")
+    zero_stage = state_dicts[0][OPTIMIZER_STATE_DICT][ZERO_STAGE]
+    world_size = state_dicts[0][OPTIMIZER_STATE_DICT][PARTITION_COUNT]
+
+    # For ZeRO-2 each param group can have different partition_count as data parallelism for expert
+    # parameters can be different from data parallelism for non-expert parameters. So we can just
+    # use the max of the partition_count to get the dp world_size.
+
+    if type(world_size) is list:
+        world_size = max(world_size)
+
+    if world_size != total_files:
+        raise ValueError(
+            f"Expected {world_size} of '*_optim_states.pt' under '{ds_checkpoint_dir}' but found {total_files} files. "
+            "Possibly due to an overwrite of an old checkpoint, or a checkpoint didn't get saved by one or more processes."
+        )
+
+    # the groups are named differently in each stage
+    if zero_stage == 2:
+        fp32_groups_key = SINGLE_PARTITION_OF_FP32_GROUPS
+    elif zero_stage == 3:
+        fp32_groups_key = FP32_FLAT_GROUPS
+    else:
+        raise ValueError(f"unknown zero stage {zero_stage}")
+
+    if zero_stage == 2:
+        fp32_flat_groups = [state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key] for i in range(len(state_dicts))]
+    elif zero_stage == 3:
+        # if there is more than one param group, there will be multiple flattened tensors - one
+        # flattened tensor per group - for simplicity merge them into a single tensor
+        #
+        # XXX: could make the script more memory efficient for when there are multiple groups - it
+        # will require matching the sub-lists of param_shapes for each param group flattened tensor
+
+        fp32_flat_groups = [
+            torch.cat(state_dicts[i][OPTIMIZER_STATE_DICT][fp32_groups_key], 0) for i in range(len(state_dicts))
+        ]
+
+    return zero_stage, world_size, fp32_flat_groups
+
+
+def _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir):
+    """
+    Returns fp32 state_dict reconstructed from ds checkpoint
+
+    Args:
+        - ``ds_checkpoint_dir``: path to the deepspeed checkpoint folder (where the optimizer files are)
+
+    """
+    print(f"Processing zero checkpoint '{ds_checkpoint_dir}'")
+
+    optim_files = get_optim_files(ds_checkpoint_dir)
+    zero_stage, world_size, fp32_flat_groups = parse_optim_states(optim_files, ds_checkpoint_dir)
+    print(f"Detected checkpoint of type zero stage {zero_stage}, world_size: {world_size}")
+
+    model_files = get_model_state_files(ds_checkpoint_dir)
+
+    zero_model_states = parse_model_states(model_files)
+    print(f'Parsing checkpoint created by deepspeed=={zero_model_states[0].ds_version}')
+
+    if zero_stage == 2:
+        return _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+    elif zero_stage == 3:
+        return _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states)
+
+
+def _zero2_merge_frozen_params(state_dict, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+    frozen_param_fragments = zero_model_states[0].frozen_param_fragments
+
+    if debug:
+        num_elem = sum(s.numel() for s in frozen_param_shapes.values())
+        print(f'rank 0: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in frozen_param_fragments.values()])
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        state_dict[name] = frozen_param_fragments[name]
+
+        if debug:
+            print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+
+    # Reconstruction protocol:
+    #
+    # XXX: document this
+
+    if debug:
+        for i in range(world_size):
+            for j in range(len(fp32_flat_groups[0])):
+                print(f"{FP32_FLAT_GROUPS}[{i}][{j}].shape={fp32_flat_groups[i][j].shape}")
+
+    # XXX: memory usage doubles here (zero2)
+    num_param_groups = len(fp32_flat_groups[0])
+    merged_single_partition_of_fp32_groups = []
+    for i in range(num_param_groups):
+        merged_partitions = [sd[i] for sd in fp32_flat_groups]
+        full_single_fp32_vector = torch.cat(merged_partitions, 0)
+        merged_single_partition_of_fp32_groups.append(full_single_fp32_vector)
+    avail_numel = sum(
+        [full_single_fp32_vector.numel() for full_single_fp32_vector in merged_single_partition_of_fp32_groups])
+
+    if debug:
+        wanted_params = sum([len(shapes) for shapes in param_shapes])
+        wanted_numel = sum([sum(shape.numel() for shape in shapes.values()) for shapes in param_shapes])
+        # not asserting if there is a mismatch due to possible padding
+        print(f"Have {avail_numel} numels to process.")
+        print(f"Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    total_numel = 0
+    total_params = 0
+    for shapes, full_single_fp32_vector in zip(param_shapes, merged_single_partition_of_fp32_groups):
+        offset = 0
+        avail_numel = full_single_fp32_vector.numel()
+        for name, shape in shapes.items():
+
+            unpartitioned_numel = shape.numel()
+            total_numel += unpartitioned_numel
+            total_params += 1
+
+            if debug:
+                print(f"{name} full shape: {shape} unpartitioned numel {unpartitioned_numel} ")
+            state_dict[name] = full_single_fp32_vector.narrow(0, offset, unpartitioned_numel).view(shape)
+            offset += unpartitioned_numel
+
+        # Z2 started to align to 2*world_size to improve nccl performance. Therefore both offset and
+        # avail_numel can differ by anywhere between 0..2*world_size. Due to two unrelated complex
+        # paddings performed in the code it's almost impossible to predict the exact numbers w/o the
+        # live optimizer object, so we are checking that the numbers are within the right range
+        align_to = 2 * world_size
+
+        def zero2_align(x):
+            return align_to * math.ceil(x / align_to)
+
+        if debug:
+            print(f"original offset={offset}, avail_numel={avail_numel}")
+
+        offset = zero2_align(offset)
+        avail_numel = zero2_align(avail_numel)
+
+        if debug:
+            print(f"aligned  offset={offset}, avail_numel={avail_numel}")
+
+        # Sanity check
+        if offset != avail_numel:
+            raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero2_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero2_merge_frozen_params(state_dict, zero_model_states)
+
+    _zero2_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def zero3_partitioned_param_info(unpartitioned_numel, world_size):
+    remainder = unpartitioned_numel % world_size
+    padding_numel = (world_size - remainder) if remainder else 0
+    partitioned_numel = math.ceil(unpartitioned_numel / world_size)
+    return partitioned_numel, padding_numel
+
+
+def _zero3_merge_frozen_params(state_dict, world_size, zero_model_states):
+    if zero_model_states[0].frozen_param_shapes is None or len(zero_model_states[0].frozen_param_shapes) == 0:
+        return
+
+    if debug:
+        for i in range(world_size):
+            num_elem = sum(s.numel() for s in zero_model_states[i].frozen_param_fragments.values())
+            print(f'rank {i}: {FROZEN_PARAM_SHAPES}.numel = {num_elem}')
+
+        frozen_param_shapes = zero_model_states[0].frozen_param_shapes
+        wanted_params = len(frozen_param_shapes)
+        wanted_numel = sum(s.numel() for s in frozen_param_shapes.values())
+        avail_numel = sum([p.numel() for p in zero_model_states[0].frozen_param_fragments.values()]) * world_size
+        print(f'Frozen params: Have {avail_numel} numels to process.')
+        print(f'Frozen params: Need {wanted_numel} numels in {wanted_params} params')
+
+    total_params = 0
+    total_numel = 0
+    for name, shape in zero_model_states[0].frozen_param_shapes.items():
+        total_params += 1
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+
+        param_frags = tuple(model_state.frozen_param_fragments[name] for model_state in zero_model_states)
+        state_dict[name] = torch.cat(param_frags, 0).narrow(0, 0, unpartitioned_numel).view(shape)
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Frozen params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+    print(f"Reconstructed Frozen fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states):
+    param_shapes = zero_model_states[0].param_shapes
+    avail_numel = fp32_flat_groups[0].numel() * world_size
+    # Reconstruction protocol: For zero3 we need to zip the partitions together at boundary of each
+    # param, re-consolidating each param, while dealing with padding if any
+
+    # merge list of dicts, preserving order
+    param_shapes = {k: v for d in param_shapes for k, v in d.items()}
+
+    if debug:
+        for i in range(world_size):
+            print(f"{FP32_FLAT_GROUPS}[{i}].shape={fp32_flat_groups[i].shape}")
+
+        wanted_params = len(param_shapes)
+        wanted_numel = sum(shape.numel() for shape in param_shapes.values())
+        # not asserting if there is a mismatch due to possible padding
+        avail_numel = fp32_flat_groups[0].numel() * world_size
+        print(f"Trainable params: Have {avail_numel} numels to process.")
+        print(f"Trainable params: Need {wanted_numel} numels in {wanted_params} params.")
+
+    # params
+    # XXX: for huge models that can't fit into the host's RAM we will have to recode this to support
+    # out-of-core computing solution
+    offset = 0
+    total_numel = 0
+    total_params = 0
+    for name, shape in param_shapes.items():
+
+        unpartitioned_numel = shape.numel()
+        total_numel += unpartitioned_numel
+        total_params += 1
+
+        partitioned_numel, partitioned_padding_numel = zero3_partitioned_param_info(unpartitioned_numel, world_size)
+
+        if debug:
+            print(
+                f"Trainable params: {total_params} {name} full shape: {shape} partition0 numel={partitioned_numel} partitioned_padding_numel={partitioned_padding_numel}"
+            )
+
+        # XXX: memory usage doubles here
+        state_dict[name] = torch.cat(
+            tuple(fp32_flat_groups[i].narrow(0, offset, partitioned_numel) for i in range(world_size)),
+            0).narrow(0, 0, unpartitioned_numel).view(shape)
+        offset += partitioned_numel
+
+    offset *= world_size
+
+    # Sanity check
+    if offset != avail_numel:
+        raise ValueError(f"consumed {offset} numels out of {avail_numel} - something is wrong")
+
+    print(f"Reconstructed Trainable fp32 state dict with {total_params} params {total_numel} elements")
+
+
+def _get_fp32_state_dict_from_zero3_checkpoint(world_size, fp32_flat_groups, zero_model_states):
+    state_dict = OrderedDict()
+
+    # buffers
+    buffers = zero_model_states[0].buffers
+    state_dict.update(buffers)
+    if debug:
+        print(f"added {len(buffers)} buffers")
+
+    _zero3_merge_frozen_params(state_dict, world_size, zero_model_states)
+
+    _zero3_merge_trainable_params(state_dict, world_size, fp32_flat_groups, zero_model_states)
+
+    # recover shared parameters
+    for pair in zero_model_states[0].shared_params:
+        if pair[1] in state_dict:
+            state_dict[pair[0]] = state_dict[pair[1]]
+
+    return state_dict
+
+
+def get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated state_dict that can be loaded with
+    ``load_state_dict()`` and used for training without DeepSpeed or shared with others, for example
+    via a model hub.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in 'latest' file. e.g., ``global_step14``
+
+    Returns:
+        - pytorch ``state_dict``
+
+    Note: this approach may not work if your application doesn't have sufficient free CPU memory and
+    you may need to use the offline approach using the ``zero_to_fp32.py`` script that is saved with
+    the checkpoint.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import get_fp32_state_dict_from_zero_checkpoint
+        # do the training and checkpoint saving
+        state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir) # already on cpu
+        model = model.cpu() # move to cpu
+        model.load_state_dict(state_dict)
+        # submit to model hub or save the model to share with others
+
+    In this example the ``model`` will no longer be usable in the deepspeed context of the same
+    application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    If you want it all done for you, use ``load_state_dict_from_zero_checkpoint`` instead.
+
+    """
+    if tag is None:
+        latest_path = os.path.join(checkpoint_dir, 'latest')
+        if os.path.isfile(latest_path):
+            with open(latest_path, 'r') as fd:
+                tag = fd.read().strip()
+        else:
+            raise ValueError(f"Unable to find 'latest' file at {latest_path}")
+
+    ds_checkpoint_dir = os.path.join(checkpoint_dir, tag)
+
+    if not os.path.isdir(ds_checkpoint_dir):
+        raise FileNotFoundError(f"Directory '{ds_checkpoint_dir}' doesn't exist")
+
+    return _get_fp32_state_dict_from_zero_checkpoint(ds_checkpoint_dir)
+
+
+def convert_zero_checkpoint_to_fp32_state_dict(checkpoint_dir, output_file, tag=None):
+    """
+    Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict`` file that can be
+    loaded with ``torch.load(file)`` + ``load_state_dict()`` and used for training without DeepSpeed.
+
+    Args:
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``output_file``: path to the pytorch fp32 state_dict output file (e.g. path/pytorch_model.bin)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+    """
+
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+    print(f"Saving fp32 state dict to {output_file}")
+    torch.save(state_dict, output_file)
+
+
+def load_state_dict_from_zero_checkpoint(model, checkpoint_dir, tag=None):
+    """
+    1. Put the provided model to cpu
+    2. Convert ZeRO 2 or 3 checkpoint into a single fp32 consolidated ``state_dict``
+    3. Load it into the provided model
+
+    Args:
+        - ``model``: the model object to update
+        - ``checkpoint_dir``: path to the desired checkpoint folder. (one that contains the tag-folder, like ``global_step14``)
+        - ``tag``: checkpoint tag used as a unique identifier for checkpoint. If not provided will attempt to load tag in the file named ``latest`` in the checkpoint folder, e.g., ``global_step14``
+
+    Returns:
+        - ``model`: modified model
+
+    Make sure you have plenty of CPU memory available before you call this function. If you don't
+    have enough use the ``zero_to_fp32.py`` utility to do the conversion. You will find it
+    conveniently placed for you in the checkpoint folder.
+
+    A typical usage might be ::
+
+        from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
+        model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
+        # submit to model hub or save the model to share with others
+
+    Note, that once this was run, the ``model`` will no longer be usable in the deepspeed context
+    of the same application. i.e. you will need to re-initialize the deepspeed engine, since
+    ``model.load_state_dict(state_dict)`` will remove all the deepspeed magic from it.
+
+    """
+    logger.info(f"Extracting fp32 weights")
+    state_dict = get_fp32_state_dict_from_zero_checkpoint(checkpoint_dir, tag)
+
+    logger.info(f"Overwriting model with fp32 weights")
+    model = model.cpu()
+    model.load_state_dict(state_dict, strict=False)
+
+    return model
+
+
+if __name__ == "__main__":
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("checkpoint_dir",
+                        type=str,
+                        help="path to the desired checkpoint folder, e.g., path/checkpoint-12")
+    parser.add_argument(
+        "output_file",
+        type=str,
+        help="path to the pytorch fp32 state_dict output file (e.g. path/checkpoint-12/pytorch_model.bin)")
+    parser.add_argument("-d", "--debug", action='store_true', help="enable debug")
+    args = parser.parse_args()
+
+    debug = args.debug
+
+    convert_zero_checkpoint_to_fp32_state_dict(args.checkpoint_dir, args.output_file)

config.json

@@ -0,0 +1,26 @@
+{
+  "_name_or_path": "/data/searchgpt/yq/GoGPT/outputs-pt-v1-7b-llama2/ckpt",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "num_key_value_heads": 32,
+  "pad_token_id": 0,
+  "pretraining_tp": 1,
+  "rms_norm_eps": 1e-05,
+  "rope_scaling": null,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "4.29.1",
+  "use_cache": false,
+  "vocab_size": 68419
+}

eval_results.json

@@ -0,0 +1,10 @@
+{
+    "epoch": 1.0,
+    "eval_accuracy": 0.3399546485260771,
+    "eval_loss": 4.18359375,
+    "eval_runtime": 6.4986,
+    "eval_samples": 35,
+    "eval_samples_per_second": 5.386,
+    "eval_steps_per_second": 0.308,
+    "perplexity": 65.60118435636834
+}

generation_config.json

@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 32000,
+  "temperature": 0.9,
+  "top_p": 0.6,
+  "transformers_version": "4.29.1"
+}

pytorch_model-00001-of-00002.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e27a3fe69c3ac40f851ec16e6c0c29431d3ba07234fca4bec97ba11215aef46c
+size 10531361877

pytorch_model-00002-of-00002.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f40b716a3e179d24dc15bda48d5351ddb0461de378fcfd377f90b72b35afe9f
+size 4663254577

pytorch_model.bin.index.json

@@ -0,0 +1,330 @@
+{
+  "metadata": {
+    "total_size": 14073524224
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model-00002-of-00002.bin",
+    "model.embed_tokens.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.30.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.input_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00002.bin",
+    "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00002.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00002.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00002.bin",
+    "model.norm.weight": "pytorch_model-00002-of-00002.bin"
+  }
+}

runs/Jul19_14-52-29_715436/1689749887.7189374/events.out.tfevents.1689749887.715436.72469.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdb3cad54708ca4ea5586145e005df66d6c84489cff128cf1c816fe6f416512a
+size 6251

runs/Jul19_14-52-29_715436/events.out.tfevents.1689749887.715436.72469.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8815ad6acd0286a0dc57ad5d7f01253581be85c195ac1dae505c28e353513812
+size 4225

runs/Jul19_14-59-01_715436/1689750342.6405456/events.out.tfevents.1689750342.715436.75291.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:824f48312c8b4ec9a8cbfc3cc0e029653beb01dfe6d6cf578f4e7aaf0b3c8fae
+size 6251

runs/Jul19_14-59-01_715436/events.out.tfevents.1689750342.715436.75291.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a275337bac363e0497d15ec03d8b4957b4a761f3045808c76fa101552f6a70b4
+size 4995

runs/Jul19_15-22-48_715436/1689751771.9245906/events.out.tfevents.1689751771.715436.80001.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c67b1bcfb983c36a142d0234c432c10bba6e2156585bb778fcfbed1f6c67d2e4
+size 6251

runs/Jul19_15-22-48_715436/events.out.tfevents.1689751771.715436.80001.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b264be3936e7ff1192446687c427473497e783f5777a6fde381d01ebccfc7e9
+size 42475

runs/Jul20_03-05-51_715436/1689793875.322509/events.out.tfevents.1689793875.715436.71505.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86bf4a00156db71e497fd9ab4779bf97c0dfa3bb64e9311574ea725b353ea87e
+size 6251

runs/Jul20_03-05-51_715436/events.out.tfevents.1689793875.715436.71505.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b5b12266ab3267dc906c4e0b6fecd8c43ece594b0e7956348d9b3eccac786c22
+size 112320

runs/Jul20_03-05-51_715436/events.out.tfevents.1689902211.715436.71505.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dfa5cf9724eb82d20e7b1fac4d2384bf340bd2525972bb89b296e130344c4f0
+size 411

special_tokens_map.json

@@ -0,0 +1,23 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd5ab2c18ed07a14f3aa55518dcf08bbee4fe86c9423e86ba61f60a82ab31fa7
+size 1077901

tokenizer_config.json

@@ -0,0 +1,35 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": null,
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "trust_remote_code": true,
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "use_fast": false
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 1.0,
+    "train_loss": 3.7094925158997714,
+    "train_runtime": 108243.5499,
+    "train_samples": 311198,
+    "train_samples_per_second": 2.875,
+    "train_steps_per_second": 0.045
+}

trainer_state.json

@@ -0,0 +1,3820 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998971722365039,
+  "global_step": 4862,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0,
+      "loss": 11.981,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9970117109141705e-05,
+      "loss": 12.0789,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.331893267009234e-05,
+      "loss": 10.7133,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00011407670594843083,
+      "loss": 8.7339,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00012721122651399258,
+      "loss": 8.301,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00013684136855727938,
+      "loss": 8.1964,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "eval_accuracy": 0.10310832025117739,
+      "eval_loss": 8.203125,
+      "eval_runtime": 6.5764,
+      "eval_samples_per_second": 5.322,
+      "eval_steps_per_second": 0.304,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014444862339428802,
+      "loss": 8.0553,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015073705430110066,
+      "loss": 7.9436,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015609707636042195,
+      "loss": 7.8368,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016076788727202945,
+      "loss": 7.7333,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016490670495758757,
+      "loss": 7.6139,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "eval_accuracy": 0.12459794174079888,
+      "eval_loss": 7.81640625,
+      "eval_runtime": 6.6059,
+      "eval_samples_per_second": 5.298,
+      "eval_steps_per_second": 0.303,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001686224178807056,
+      "loss": 7.4892,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000171993565594773,
+      "loss": 7.3256,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017507866443784335,
+      "loss": 7.1827,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001779224840062419,
+      "loss": 6.9698,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018056004207494319,
+      "loss": 6.8162,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "eval_accuracy": 0.1567556253270539,
+      "eval_loss": 7.08203125,
+      "eval_runtime": 6.5908,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018301924610008189,
+      "loss": 6.6293,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018532269677939782,
+      "loss": 6.4114,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018748895370481112,
+      "loss": 6.2911,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00018953344483335556,
+      "loss": 6.1047,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019146913367833817,
+      "loss": 5.9957,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "eval_accuracy": 0.19811616954474098,
+      "eval_loss": 6.4296875,
+      "eval_runtime": 6.5897,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.304,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019330701776944063,
+      "loss": 5.8281,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019505650713185044,
+      "loss": 5.6927,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019672571585424665,
+      "loss": 5.5564,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019832168964685297,
+      "loss": 5.3813,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998505855457085,
+      "loss": 5.2496,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "eval_accuracy": 0.24055817198674342,
+      "eval_loss": 5.8203125,
+      "eval_runtime": 6.6013,
+      "eval_samples_per_second": 5.302,
+      "eval_steps_per_second": 0.303,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996535296665223,
+      "loss": 5.13,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001992204417496752,
+      "loss": 5.0354,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019878735383282807,
+      "loss": 4.9021,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019835426591598097,
+      "loss": 4.8181,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019792117799913384,
+      "loss": 4.6993,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "eval_accuracy": 0.2641758241758242,
+      "eval_loss": 5.41796875,
+      "eval_runtime": 6.6009,
+      "eval_samples_per_second": 5.302,
+      "eval_steps_per_second": 0.303,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001974880900822867,
+      "loss": 4.6761,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019705500216543958,
+      "loss": 4.5908,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019662191424859245,
+      "loss": 4.5301,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019618882633174535,
+      "loss": 4.4729,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019575573841489822,
+      "loss": 4.3928,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "eval_accuracy": 0.27927437641723357,
+      "eval_loss": 5.14453125,
+      "eval_runtime": 6.6101,
+      "eval_samples_per_second": 5.295,
+      "eval_steps_per_second": 0.303,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019532265049805112,
+      "loss": 4.4083,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000194889562581204,
+      "loss": 4.336,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001944564746643569,
+      "loss": 4.2714,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019402338674750976,
+      "loss": 4.2124,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019359029883066263,
+      "loss": 4.2395,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "eval_accuracy": 0.2894470608756323,
+      "eval_loss": 4.96875,
+      "eval_runtime": 6.5945,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001931572109138155,
+      "loss": 4.1867,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001927241229969684,
+      "loss": 4.1687,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019229103508012127,
+      "loss": 4.1027,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019185794716327414,
+      "loss": 4.1233,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019142485924642704,
+      "loss": 4.0781,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "eval_accuracy": 0.29638932496075354,
+      "eval_loss": 4.8515625,
+      "eval_runtime": 6.5996,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001909917713295799,
+      "loss": 4.0855,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001905586834127328,
+      "loss": 4.0859,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019012559549588568,
+      "loss": 4.0124,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00018969250757903855,
+      "loss": 4.0151,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00018925941966219142,
+      "loss": 4.0409,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "eval_accuracy": 0.30181057038199893,
+      "eval_loss": 4.76953125,
+      "eval_runtime": 6.6117,
+      "eval_samples_per_second": 5.294,
+      "eval_steps_per_second": 0.302,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00018882633174534431,
+      "loss": 3.9912,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018839324382849718,
+      "loss": 3.9383,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018796015591165008,
+      "loss": 3.9764,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018752706799480295,
+      "loss": 3.9809,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00018709398007795585,
+      "loss": 3.9178,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "eval_accuracy": 0.30575963718820864,
+      "eval_loss": 4.703125,
+      "eval_runtime": 6.6265,
+      "eval_samples_per_second": 5.282,
+      "eval_steps_per_second": 0.302,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018666089216110872,
+      "loss": 3.9073,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001862278042442616,
+      "loss": 3.9459,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018579471632741446,
+      "loss": 3.9535,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018536162841056733,
+      "loss": 3.8982,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00018492854049372023,
+      "loss": 3.834,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "eval_accuracy": 0.3082574568288854,
+      "eval_loss": 4.65625,
+      "eval_runtime": 6.6135,
+      "eval_samples_per_second": 5.292,
+      "eval_steps_per_second": 0.302,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001844954525768731,
+      "loss": 3.8759,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000184062364660026,
+      "loss": 3.8835,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00018362927674317887,
+      "loss": 3.9003,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00018319618882633177,
+      "loss": 3.8538,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00018276310090948464,
+      "loss": 3.8316,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "eval_accuracy": 0.30981336124193265,
+      "eval_loss": 4.625,
+      "eval_runtime": 6.6039,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018233001299263754,
+      "loss": 3.9085,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018189692507579038,
+      "loss": 3.7825,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018146383715894328,
+      "loss": 3.824,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018103074924209615,
+      "loss": 3.8457,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00018059766132524902,
+      "loss": 3.8197,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "eval_accuracy": 0.3116902145473574,
+      "eval_loss": 4.59765625,
+      "eval_runtime": 6.6155,
+      "eval_samples_per_second": 5.291,
+      "eval_steps_per_second": 0.302,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00018016457340840192,
+      "loss": 3.79,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017973148549155479,
+      "loss": 3.7907,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017929839757470768,
+      "loss": 3.7797,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017886530965786055,
+      "loss": 3.7533,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017843222174101345,
+      "loss": 3.7464,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "eval_accuracy": 0.31469038897610324,
+      "eval_loss": 4.5625,
+      "eval_runtime": 6.5988,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017799913382416632,
+      "loss": 3.7347,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001775660459073192,
+      "loss": 3.7917,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017713295799047206,
+      "loss": 3.8106,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017669987007362496,
+      "loss": 3.7289,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017626678215677783,
+      "loss": 3.767,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "eval_accuracy": 0.3163718820861678,
+      "eval_loss": 4.5390625,
+      "eval_runtime": 6.604,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017583369423993073,
+      "loss": 3.7362,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001754006063230836,
+      "loss": 3.7474,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017496751840623647,
+      "loss": 3.7485,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017453443048938937,
+      "loss": 3.7341,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00017410134257254224,
+      "loss": 3.7511,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "eval_accuracy": 0.3172649572649573,
+      "eval_loss": 4.515625,
+      "eval_runtime": 6.6093,
+      "eval_samples_per_second": 5.296,
+      "eval_steps_per_second": 0.303,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001736682546556951,
+      "loss": 3.6862,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017323516673884798,
+      "loss": 3.6411,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017280207882200088,
+      "loss": 3.7181,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00017236899090515375,
+      "loss": 3.6471,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017193590298830665,
+      "loss": 3.7166,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "eval_accuracy": 0.3188278388278388,
+      "eval_loss": 4.4921875,
+      "eval_runtime": 6.6113,
+      "eval_samples_per_second": 5.294,
+      "eval_steps_per_second": 0.303,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017150281507145952,
+      "loss": 3.6869,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017106972715461241,
+      "loss": 3.6728,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017063663923776528,
+      "loss": 3.7705,
+      "step": 930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00017020355132091815,
+      "loss": 3.6728,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00016977046340407103,
+      "loss": 3.6908,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "eval_accuracy": 0.3190441304727019,
+      "eval_loss": 4.48046875,
+      "eval_runtime": 6.5961,
+      "eval_samples_per_second": 5.306,
+      "eval_steps_per_second": 0.303,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001693373754872239,
+      "loss": 3.5988,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001689042875703768,
+      "loss": 3.6713,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00016847119965352966,
+      "loss": 3.7165,
+      "step": 980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00016803811173668256,
+      "loss": 3.7098,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016760502381983543,
+      "loss": 3.617,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "eval_accuracy": 0.3208442351299494,
+      "eval_loss": 4.46484375,
+      "eval_runtime": 6.5949,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016717193590298833,
+      "loss": 3.6629,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001667388479861412,
+      "loss": 3.684,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016630576006929407,
+      "loss": 3.6877,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00016587267215244694,
+      "loss": 3.6274,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00016543958423559984,
+      "loss": 3.6734,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "eval_accuracy": 0.3213535670678528,
+      "eval_loss": 4.453125,
+      "eval_runtime": 6.5974,
+      "eval_samples_per_second": 5.305,
+      "eval_steps_per_second": 0.303,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001650064963187527,
+      "loss": 3.6895,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001645734084019056,
+      "loss": 3.6662,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00016414032048505848,
+      "loss": 3.6152,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00016370723256821135,
+      "loss": 3.6531,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016327414465136425,
+      "loss": 3.6916,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "eval_accuracy": 0.32398395255538115,
+      "eval_loss": 4.43359375,
+      "eval_runtime": 6.5897,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.304,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016284105673451712,
+      "loss": 3.6961,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016240796881767,
+      "loss": 3.5811,
+      "step": 1120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016197488090082286,
+      "loss": 3.6161,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00016154179298397576,
+      "loss": 3.6305,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016110870506712863,
+      "loss": 3.629,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "eval_accuracy": 0.32400488400488403,
+      "eval_loss": 4.421875,
+      "eval_runtime": 6.5985,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016067561715028152,
+      "loss": 3.6467,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001602425292334344,
+      "loss": 3.6573,
+      "step": 1170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001598094413165873,
+      "loss": 3.6372,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00015937635339974016,
+      "loss": 3.6369,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00015894326548289303,
+      "loss": 3.6001,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "eval_accuracy": 0.3247165532879819,
+      "eval_loss": 4.4140625,
+      "eval_runtime": 6.602,
+      "eval_samples_per_second": 5.301,
+      "eval_steps_per_second": 0.303,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001585101775660459,
+      "loss": 3.5843,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00015807708964919877,
+      "loss": 3.6407,
+      "step": 1220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00015764400173235167,
+      "loss": 3.6413,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00015721091381550454,
+      "loss": 3.5963,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00015677782589865744,
+      "loss": 3.6053,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "eval_accuracy": 0.32607709750566893,
+      "eval_loss": 4.40234375,
+      "eval_runtime": 6.5982,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001563447379818103,
+      "loss": 3.6049,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001559116500649632,
+      "loss": 3.6112,
+      "step": 1270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00015547856214811608,
+      "loss": 3.5872,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015504547423126895,
+      "loss": 3.6328,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015461238631442182,
+      "loss": 3.5803,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "eval_accuracy": 0.32540729112157685,
+      "eval_loss": 4.390625,
+      "eval_runtime": 6.5995,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015417929839757472,
+      "loss": 3.6153,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001537462104807276,
+      "loss": 3.5493,
+      "step": 1320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00015331312256388049,
+      "loss": 3.5854,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015288003464703336,
+      "loss": 3.6029,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015244694673018623,
+      "loss": 3.5886,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "eval_accuracy": 0.32738880167451595,
+      "eval_loss": 4.37890625,
+      "eval_runtime": 6.5931,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015201385881333913,
+      "loss": 3.5496,
+      "step": 1360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000151580770896492,
+      "loss": 3.557,
+      "step": 1370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00015114768297964487,
+      "loss": 3.5647,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015071459506279774,
+      "loss": 3.5912,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00015028150714595063,
+      "loss": 3.5033,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "eval_accuracy": 0.3287981859410431,
+      "eval_loss": 4.3671875,
+      "eval_runtime": 6.593,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001498484192291035,
+      "loss": 3.5814,
+      "step": 1410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001494153313122564,
+      "loss": 3.5834,
+      "step": 1420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00014898224339540927,
+      "loss": 3.5661,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014854915547856217,
+      "loss": 3.5844,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014811606756171504,
+      "loss": 3.58,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "eval_accuracy": 0.3283656026513169,
+      "eval_loss": 4.36328125,
+      "eval_runtime": 6.6004,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001476829796448679,
+      "loss": 3.5874,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014724989172802078,
+      "loss": 3.5736,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00014681680381117365,
+      "loss": 3.5659,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014638371589432655,
+      "loss": 3.5632,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014595062797747942,
+      "loss": 3.4966,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "eval_accuracy": 0.32832373975231116,
+      "eval_loss": 4.3515625,
+      "eval_runtime": 6.592,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00014551754006063232,
+      "loss": 3.5236,
+      "step": 1510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001450844521437852,
+      "loss": 3.5277,
+      "step": 1520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001446513642269381,
+      "loss": 3.5237,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014421827631009096,
+      "loss": 3.5719,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014378518839324383,
+      "loss": 3.5411,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "eval_accuracy": 0.32884004884004886,
+      "eval_loss": 4.3515625,
+      "eval_runtime": 6.6086,
+      "eval_samples_per_second": 5.296,
+      "eval_steps_per_second": 0.303,
+      "step": 1550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001433521004763967,
+      "loss": 3.5287,
+      "step": 1560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001429190125595496,
+      "loss": 3.5965,
+      "step": 1570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00014248592464270247,
+      "loss": 3.5435,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014205283672585536,
+      "loss": 3.5536,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00014161974880900824,
+      "loss": 3.527,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "eval_accuracy": 0.33027036455607883,
+      "eval_loss": 4.33984375,
+      "eval_runtime": 6.5917,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 1600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001411866608921611,
+      "loss": 3.5765,
+      "step": 1610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000140753572975314,
+      "loss": 3.5882,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00014032048505846687,
+      "loss": 3.5135,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013988739714161974,
+      "loss": 3.4924,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013945430922477262,
+      "loss": 3.6018,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "eval_accuracy": 0.3299284842141985,
+      "eval_loss": 4.33203125,
+      "eval_runtime": 6.5992,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 1650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001390212213079255,
+      "loss": 3.5646,
+      "step": 1660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013858813339107838,
+      "loss": 3.5164,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013815504547423128,
+      "loss": 3.5433,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013772195755738415,
+      "loss": 3.4929,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013728886964053705,
+      "loss": 3.4802,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "eval_accuracy": 0.3301238444095587,
+      "eval_loss": 4.32421875,
+      "eval_runtime": 6.5899,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.303,
+      "step": 1700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013685578172368992,
+      "loss": 3.5211,
+      "step": 1710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013642269380684282,
+      "loss": 3.5424,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013598960588999566,
+      "loss": 3.468,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013555651797314853,
+      "loss": 3.5342,
+      "step": 1740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013512343005630143,
+      "loss": 3.4375,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "eval_accuracy": 0.330584336298622,
+      "eval_loss": 4.32421875,
+      "eval_runtime": 6.5928,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 1750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001346903421394543,
+      "loss": 3.5239,
+      "step": 1760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001342572542226072,
+      "loss": 3.5176,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013382416630576007,
+      "loss": 3.4992,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013339107838891297,
+      "loss": 3.457,
+      "step": 1790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013295799047206584,
+      "loss": 3.4873,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "eval_accuracy": 0.33109366823652536,
+      "eval_loss": 4.3203125,
+      "eval_runtime": 6.6125,
+      "eval_samples_per_second": 5.293,
+      "eval_steps_per_second": 0.302,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013252490255521873,
+      "loss": 3.5021,
+      "step": 1810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001320918146383716,
+      "loss": 3.4871,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013165872672152448,
+      "loss": 3.47,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013122563880467735,
+      "loss": 3.4462,
+      "step": 1840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013079255088783024,
+      "loss": 3.435,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "eval_accuracy": 0.3309401709401709,
+      "eval_loss": 4.3125,
+      "eval_runtime": 6.5916,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 1850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013035946297098311,
+      "loss": 3.4994,
+      "step": 1860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00012992637505413598,
+      "loss": 3.5665,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012949328713728888,
+      "loss": 3.5552,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012906019922044175,
+      "loss": 3.5092,
+      "step": 1890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012862711130359465,
+      "loss": 3.4335,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "eval_accuracy": 0.3317634746206175,
+      "eval_loss": 4.3046875,
+      "eval_runtime": 6.6183,
+      "eval_samples_per_second": 5.288,
+      "eval_steps_per_second": 0.302,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012819402338674752,
+      "loss": 3.5154,
+      "step": 1910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001277609354699004,
+      "loss": 3.5207,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012732784755305326,
+      "loss": 3.4869,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012689475963620616,
+      "loss": 3.4773,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012646167171935903,
+      "loss": 3.4595,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "eval_accuracy": 0.3325239839525554,
+      "eval_loss": 4.296875,
+      "eval_runtime": 6.5968,
+      "eval_samples_per_second": 5.306,
+      "eval_steps_per_second": 0.303,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012602858380251193,
+      "loss": 3.5035,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001255954958856648,
+      "loss": 3.4692,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001251624079688177,
+      "loss": 3.4712,
+      "step": 1980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012472932005197057,
+      "loss": 3.4558,
+      "step": 1990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012429623213512344,
+      "loss": 3.4937,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "eval_accuracy": 0.3319448805163091,
+      "eval_loss": 4.29296875,
+      "eval_runtime": 6.5908,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001238631442182763,
+      "loss": 3.5093,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012343005630142918,
+      "loss": 3.4941,
+      "step": 2020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012299696838458208,
+      "loss": 3.4706,
+      "step": 2030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012256388046773495,
+      "loss": 3.5106,
+      "step": 2040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012213079255088784,
+      "loss": 3.4959,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "eval_accuracy": 0.3324123495552067,
+      "eval_loss": 4.28515625,
+      "eval_runtime": 6.6003,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012169770463404072,
+      "loss": 3.5218,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001212646167171936,
+      "loss": 3.4813,
+      "step": 2070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012083152880034647,
+      "loss": 3.5107,
+      "step": 2080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012039844088349937,
+      "loss": 3.4568,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011996535296665224,
+      "loss": 3.4987,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "eval_accuracy": 0.3331658817373103,
+      "eval_loss": 4.28515625,
+      "eval_runtime": 6.6107,
+      "eval_samples_per_second": 5.294,
+      "eval_steps_per_second": 0.303,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011957557384148984,
+      "loss": 3.458,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001191424859246427,
+      "loss": 3.4656,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011870939800779559,
+      "loss": 3.4505,
+      "step": 2130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011827631009094846,
+      "loss": 3.4182,
+      "step": 2140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011784322217410136,
+      "loss": 3.4001,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "eval_accuracy": 0.3336403279260422,
+      "eval_loss": 4.28515625,
+      "eval_runtime": 6.5938,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011741013425725423,
+      "loss": 3.4289,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011697704634040711,
+      "loss": 3.4228,
+      "step": 2170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011654395842355998,
+      "loss": 3.4066,
+      "step": 2180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011611087050671288,
+      "loss": 3.4823,
+      "step": 2190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011567778258986575,
+      "loss": 3.4497,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "eval_accuracy": 0.3340380254665969,
+      "eval_loss": 4.28125,
+      "eval_runtime": 6.606,
+      "eval_samples_per_second": 5.298,
+      "eval_steps_per_second": 0.303,
+      "step": 2200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011524469467301864,
+      "loss": 3.4737,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001148116067561715,
+      "loss": 3.4564,
+      "step": 2220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011437851883932438,
+      "loss": 3.4865,
+      "step": 2230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011394543092247727,
+      "loss": 3.4496,
+      "step": 2240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011351234300563015,
+      "loss": 3.4068,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "eval_accuracy": 0.33329147043432755,
+      "eval_loss": 4.27734375,
+      "eval_runtime": 6.6133,
+      "eval_samples_per_second": 5.292,
+      "eval_steps_per_second": 0.302,
+      "step": 2250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011307925508878303,
+      "loss": 3.4337,
+      "step": 2260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001126461671719359,
+      "loss": 3.4338,
+      "step": 2270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001122130792550888,
+      "loss": 3.4241,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011177999133824167,
+      "loss": 3.497,
+      "step": 2290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011134690342139455,
+      "loss": 3.4634,
+      "step": 2300
+    },
+    {
+      "epoch": 0.47,
+      "eval_accuracy": 0.3339822082679226,
+      "eval_loss": 4.2734375,
+      "eval_runtime": 6.5925,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011091381550454742,
+      "loss": 3.4245,
+      "step": 2310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011048072758770032,
+      "loss": 3.4465,
+      "step": 2320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011004763967085319,
+      "loss": 3.498,
+      "step": 2330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010961455175400607,
+      "loss": 3.3637,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010918146383715895,
+      "loss": 3.4324,
+      "step": 2350
+    },
+    {
+      "epoch": 0.48,
+      "eval_accuracy": 0.333849642421071,
+      "eval_loss": 4.27734375,
+      "eval_runtime": 6.5996,
+      "eval_samples_per_second": 5.303,
+      "eval_steps_per_second": 0.303,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010874837592031182,
+      "loss": 3.3816,
+      "step": 2360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010831528800346471,
+      "loss": 3.4245,
+      "step": 2370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010788220008661758,
+      "loss": 3.4567,
+      "step": 2380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010744911216977047,
+      "loss": 3.4122,
+      "step": 2390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010701602425292334,
+      "loss": 3.4039,
+      "step": 2400
+    },
+    {
+      "epoch": 0.49,
+      "eval_accuracy": 0.3344147915576487,
+      "eval_loss": 4.265625,
+      "eval_runtime": 6.5992,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010658293633607624,
+      "loss": 3.4124,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010614984841922911,
+      "loss": 3.4309,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010571676050238199,
+      "loss": 3.4464,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010528367258553486,
+      "loss": 3.4136,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010485058466868776,
+      "loss": 3.4502,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5,
+      "eval_accuracy": 0.33453340310483165,
+      "eval_loss": 4.265625,
+      "eval_runtime": 6.7512,
+      "eval_samples_per_second": 5.184,
+      "eval_steps_per_second": 0.296,
+      "step": 2450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010441749675184063,
+      "loss": 3.4062,
+      "step": 2460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010398440883499351,
+      "loss": 3.3951,
+      "step": 2470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010355132091814638,
+      "loss": 3.4322,
+      "step": 2480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010311823300129926,
+      "loss": 3.4107,
+      "step": 2490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010268514508445215,
+      "loss": 3.4104,
+      "step": 2500
+    },
+    {
+      "epoch": 0.51,
+      "eval_accuracy": 0.33495900924472355,
+      "eval_loss": 4.2578125,
+      "eval_runtime": 6.5889,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 2500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010225205716760502,
+      "loss": 3.4181,
+      "step": 2510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010181896925075791,
+      "loss": 3.3979,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010138588133391078,
+      "loss": 3.3911,
+      "step": 2530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010095279341706368,
+      "loss": 3.4303,
+      "step": 2540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010051970550021655,
+      "loss": 3.5251,
+      "step": 2550
+    },
+    {
+      "epoch": 0.52,
+      "eval_accuracy": 0.33599860457003317,
+      "eval_loss": 4.2578125,
+      "eval_runtime": 6.5946,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 2550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010008661758336943,
+      "loss": 3.4337,
+      "step": 2560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.96535296665223e-05,
+      "loss": 3.3906,
+      "step": 2570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.922044174967519e-05,
+      "loss": 3.4714,
+      "step": 2580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.878735383282807e-05,
+      "loss": 3.4061,
+      "step": 2590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.835426591598095e-05,
+      "loss": 3.4176,
+      "step": 2600
+    },
+    {
+      "epoch": 0.53,
+      "eval_accuracy": 0.3363404849119135,
+      "eval_loss": 4.25,
+      "eval_runtime": 6.594,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 2600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.792117799913382e-05,
+      "loss": 3.37,
+      "step": 2610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.748809008228671e-05,
+      "loss": 3.3931,
+      "step": 2620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.705500216543959e-05,
+      "loss": 3.4052,
+      "step": 2630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.662191424859248e-05,
+      "loss": 3.4269,
+      "step": 2640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.618882633174535e-05,
+      "loss": 3.3795,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54,
+      "eval_accuracy": 0.3354404325832897,
+      "eval_loss": 4.25,
+      "eval_runtime": 6.6039,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 2650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.575573841489823e-05,
+      "loss": 3.4064,
+      "step": 2660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.532265049805112e-05,
+      "loss": 3.4207,
+      "step": 2670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.488956258120399e-05,
+      "loss": 3.4353,
+      "step": 2680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.445647466435687e-05,
+      "loss": 3.4497,
+      "step": 2690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.402338674750974e-05,
+      "loss": 3.3656,
+      "step": 2700
+    },
+    {
+      "epoch": 0.56,
+      "eval_accuracy": 0.33636839351125064,
+      "eval_loss": 4.25,
+      "eval_runtime": 6.5948,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 2700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.359029883066262e-05,
+      "loss": 3.3736,
+      "step": 2710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.315721091381551e-05,
+      "loss": 3.4236,
+      "step": 2720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.272412299696839e-05,
+      "loss": 3.4234,
+      "step": 2730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.229103508012126e-05,
+      "loss": 3.3849,
+      "step": 2740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.185794716327415e-05,
+      "loss": 3.3938,
+      "step": 2750
+    },
+    {
+      "epoch": 0.57,
+      "eval_accuracy": 0.33627769056340484,
+      "eval_loss": 4.24609375,
+      "eval_runtime": 6.5953,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 2750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.142485924642703e-05,
+      "loss": 3.375,
+      "step": 2760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.099177132957992e-05,
+      "loss": 3.4365,
+      "step": 2770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.055868341273279e-05,
+      "loss": 3.4068,
+      "step": 2780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.012559549588567e-05,
+      "loss": 3.4333,
+      "step": 2790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.969250757903855e-05,
+      "loss": 3.3757,
+      "step": 2800
+    },
+    {
+      "epoch": 0.58,
+      "eval_accuracy": 0.33639630211058785,
+      "eval_loss": 4.24609375,
+      "eval_runtime": 6.5987,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 2800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.925941966219143e-05,
+      "loss": 3.4111,
+      "step": 2810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.882633174534431e-05,
+      "loss": 3.4143,
+      "step": 2820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.839324382849718e-05,
+      "loss": 3.4307,
+      "step": 2830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.796015591165006e-05,
+      "loss": 3.3366,
+      "step": 2840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.752706799480295e-05,
+      "loss": 3.407,
+      "step": 2850
+    },
+    {
+      "epoch": 0.59,
+      "eval_accuracy": 0.33731728588871446,
+      "eval_loss": 4.234375,
+      "eval_runtime": 6.6038,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 2850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.709398007795583e-05,
+      "loss": 3.3506,
+      "step": 2860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.66608921611087e-05,
+      "loss": 3.429,
+      "step": 2870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.622780424426159e-05,
+      "loss": 3.338,
+      "step": 2880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.579471632741447e-05,
+      "loss": 3.4252,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.536162841056736e-05,
+      "loss": 3.3986,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6,
+      "eval_accuracy": 0.33657770800627945,
+      "eval_loss": 4.23828125,
+      "eval_runtime": 6.5874,
+      "eval_samples_per_second": 5.313,
+      "eval_steps_per_second": 0.304,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.492854049372023e-05,
+      "loss": 3.4023,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.449545257687311e-05,
+      "loss": 3.3919,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.4062364660026e-05,
+      "loss": 3.3691,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.362927674317888e-05,
+      "loss": 3.4017,
+      "step": 2940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.319618882633175e-05,
+      "loss": 3.4311,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "eval_accuracy": 0.3370940170940171,
+      "eval_loss": 4.234375,
+      "eval_runtime": 6.5922,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 2950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.276310090948462e-05,
+      "loss": 3.4224,
+      "step": 2960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.23300129926375e-05,
+      "loss": 3.3781,
+      "step": 2970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.189692507579039e-05,
+      "loss": 3.383,
+      "step": 2980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.146383715894327e-05,
+      "loss": 3.4332,
+      "step": 2990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.103074924209616e-05,
+      "loss": 3.3716,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "eval_accuracy": 0.33713587999302286,
+      "eval_loss": 4.234375,
+      "eval_runtime": 6.5875,
+      "eval_samples_per_second": 5.313,
+      "eval_steps_per_second": 0.304,
+      "step": 3000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.059766132524903e-05,
+      "loss": 3.4123,
+      "step": 3010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.016457340840191e-05,
+      "loss": 3.4181,
+      "step": 3020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.97314854915548e-05,
+      "loss": 3.3851,
+      "step": 3030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.929839757470768e-05,
+      "loss": 3.4224,
+      "step": 3040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.886530965786055e-05,
+      "loss": 3.3831,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "eval_accuracy": 0.33774986917844063,
+      "eval_loss": 4.23046875,
+      "eval_runtime": 6.5895,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.304,
+      "step": 3050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.843222174101343e-05,
+      "loss": 3.3965,
+      "step": 3060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.799913382416632e-05,
+      "loss": 3.4016,
+      "step": 3070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.756604590731919e-05,
+      "loss": 3.4047,
+      "step": 3080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.713295799047207e-05,
+      "loss": 3.3618,
+      "step": 3090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.669987007362494e-05,
+      "loss": 3.375,
+      "step": 3100
+    },
+    {
+      "epoch": 0.64,
+      "eval_accuracy": 0.3377638234781092,
+      "eval_loss": 4.23046875,
+      "eval_runtime": 6.5887,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 3100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.626678215677783e-05,
+      "loss": 3.3899,
+      "step": 3110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.583369423993071e-05,
+      "loss": 3.3723,
+      "step": 3120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.54006063230836e-05,
+      "loss": 3.381,
+      "step": 3130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.496751840623647e-05,
+      "loss": 3.3558,
+      "step": 3140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.453443048938935e-05,
+      "loss": 3.3677,
+      "step": 3150
+    },
+    {
+      "epoch": 0.65,
+      "eval_accuracy": 0.33846851561137276,
+      "eval_loss": 4.21875,
+      "eval_runtime": 6.5962,
+      "eval_samples_per_second": 5.306,
+      "eval_steps_per_second": 0.303,
+      "step": 3150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.410134257254223e-05,
+      "loss": 3.3414,
+      "step": 3160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.366825465569512e-05,
+      "loss": 3.4144,
+      "step": 3170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.323516673884799e-05,
+      "loss": 3.3503,
+      "step": 3180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.280207882200087e-05,
+      "loss": 3.3716,
+      "step": 3190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.236899090515376e-05,
+      "loss": 3.3968,
+      "step": 3200
+    },
+    {
+      "epoch": 0.66,
+      "eval_accuracy": 0.3386220129077272,
+      "eval_loss": 4.22265625,
+      "eval_runtime": 6.5929,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 3200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.193590298830663e-05,
+      "loss": 3.3485,
+      "step": 3210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.150281507145951e-05,
+      "loss": 3.4172,
+      "step": 3220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.106972715461238e-05,
+      "loss": 3.3727,
+      "step": 3230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.063663923776527e-05,
+      "loss": 3.3616,
+      "step": 3240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.020355132091815e-05,
+      "loss": 3.4069,
+      "step": 3250
+    },
+    {
+      "epoch": 0.67,
+      "eval_accuracy": 0.3380429094714809,
+      "eval_loss": 4.21875,
+      "eval_runtime": 6.5865,
+      "eval_samples_per_second": 5.314,
+      "eval_steps_per_second": 0.304,
+      "step": 3250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.977046340407103e-05,
+      "loss": 3.3583,
+      "step": 3260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.93373754872239e-05,
+      "loss": 3.3753,
+      "step": 3270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.890428757037679e-05,
+      "loss": 3.3443,
+      "step": 3280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.847119965352967e-05,
+      "loss": 3.3682,
+      "step": 3290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.803811173668256e-05,
+      "loss": 3.4192,
+      "step": 3300
+    },
+    {
+      "epoch": 0.68,
+      "eval_accuracy": 0.33877551020408164,
+      "eval_loss": 4.21484375,
+      "eval_runtime": 6.6033,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 3300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.760502381983543e-05,
+      "loss": 3.3657,
+      "step": 3310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.717193590298831e-05,
+      "loss": 3.3773,
+      "step": 3320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.67388479861412e-05,
+      "loss": 3.3604,
+      "step": 3330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.630576006929407e-05,
+      "loss": 3.404,
+      "step": 3340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.587267215244695e-05,
+      "loss": 3.3881,
+      "step": 3350
+    },
+    {
+      "epoch": 0.69,
+      "eval_accuracy": 0.33830804116518404,
+      "eval_loss": 4.21484375,
+      "eval_runtime": 6.6138,
+      "eval_samples_per_second": 5.292,
+      "eval_steps_per_second": 0.302,
+      "step": 3350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.543958423559982e-05,
+      "loss": 3.3855,
+      "step": 3360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.50064963187527e-05,
+      "loss": 3.314,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.457340840190559e-05,
+      "loss": 3.4034,
+      "step": 3380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.414032048505847e-05,
+      "loss": 3.3969,
+      "step": 3390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.370723256821134e-05,
+      "loss": 3.3858,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7,
+      "eval_accuracy": 0.33836385836385835,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.6099,
+      "eval_samples_per_second": 5.295,
+      "eval_steps_per_second": 0.303,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.327414465136423e-05,
+      "loss": 3.388,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.284105673451711e-05,
+      "loss": 3.3832,
+      "step": 3420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.240796881767e-05,
+      "loss": 3.3098,
+      "step": 3430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.197488090082287e-05,
+      "loss": 3.3794,
+      "step": 3440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.154179298397575e-05,
+      "loss": 3.3999,
+      "step": 3450
+    },
+    {
+      "epoch": 0.71,
+      "eval_accuracy": 0.33883132740275596,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.6039,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 3450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.110870506712864e-05,
+      "loss": 3.3278,
+      "step": 3460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0675617150281506e-05,
+      "loss": 3.3645,
+      "step": 3470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 6.024252923343438e-05,
+      "loss": 3.3563,
+      "step": 3480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.985275010827198e-05,
+      "loss": 3.358,
+      "step": 3490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9419662191424864e-05,
+      "loss": 3.3907,
+      "step": 3500
+    },
+    {
+      "epoch": 0.72,
+      "eval_accuracy": 0.33894296180010464,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.6054,
+      "eval_samples_per_second": 5.299,
+      "eval_steps_per_second": 0.303,
+      "step": 3500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.898657427457774e-05,
+      "loss": 3.4023,
+      "step": 3510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8553486357730626e-05,
+      "loss": 3.3519,
+      "step": 3520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.81203984408835e-05,
+      "loss": 3.3514,
+      "step": 3530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.768731052403639e-05,
+      "loss": 3.3151,
+      "step": 3540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.7254222607189265e-05,
+      "loss": 3.3929,
+      "step": 3550
+    },
+    {
+      "epoch": 0.73,
+      "eval_accuracy": 0.33893598465027036,
+      "eval_loss": 4.2109375,
+      "eval_runtime": 6.5935,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 3550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6821134690342135e-05,
+      "loss": 3.3983,
+      "step": 3560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.638804677349502e-05,
+      "loss": 3.3281,
+      "step": 3570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.59549588566479e-05,
+      "loss": 3.3654,
+      "step": 3580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.552187093980078e-05,
+      "loss": 3.379,
+      "step": 3590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.508878302295366e-05,
+      "loss": 3.3738,
+      "step": 3600
+    },
+    {
+      "epoch": 0.74,
+      "eval_accuracy": 0.33964067678353393,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.6063,
+      "eval_samples_per_second": 5.298,
+      "eval_steps_per_second": 0.303,
+      "step": 3600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.465569510610654e-05,
+      "loss": 3.3727,
+      "step": 3610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.426591598094414e-05,
+      "loss": 3.3401,
+      "step": 3620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.3832828064097017e-05,
+      "loss": 3.3583,
+      "step": 3630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.33997401472499e-05,
+      "loss": 3.3868,
+      "step": 3640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.296665223040277e-05,
+      "loss": 3.3839,
+      "step": 3650
+    },
+    {
+      "epoch": 0.75,
+      "eval_accuracy": 0.3392918192918193,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.5952,
+      "eval_samples_per_second": 5.307,
+      "eval_steps_per_second": 0.303,
+      "step": 3650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.253356431355565e-05,
+      "loss": 3.3713,
+      "step": 3660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.210047639670853e-05,
+      "loss": 3.3649,
+      "step": 3670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.166738847986141e-05,
+      "loss": 3.3836,
+      "step": 3680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.1234300563014294e-05,
+      "loss": 3.3227,
+      "step": 3690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.080121264616717e-05,
+      "loss": 3.3854,
+      "step": 3700
+    },
+    {
+      "epoch": 0.76,
+      "eval_accuracy": 0.3399546485260771,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.6101,
+      "eval_samples_per_second": 5.295,
+      "eval_steps_per_second": 0.303,
+      "step": 3700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0368124729320056e-05,
+      "loss": 3.3536,
+      "step": 3710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.993503681247293e-05,
+      "loss": 3.3943,
+      "step": 3720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.950194889562582e-05,
+      "loss": 3.3705,
+      "step": 3730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.9068860978778694e-05,
+      "loss": 3.3525,
+      "step": 3740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.863577306193157e-05,
+      "loss": 3.297,
+      "step": 3750
+    },
+    {
+      "epoch": 0.77,
+      "eval_accuracy": 0.339236002093145,
+      "eval_loss": 4.20703125,
+      "eval_runtime": 6.5942,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 3750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8202685145084456e-05,
+      "loss": 3.3305,
+      "step": 3760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.776959722823733e-05,
+      "loss": 3.3789,
+      "step": 3770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.733650931139022e-05,
+      "loss": 3.4127,
+      "step": 3780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6903421394543095e-05,
+      "loss": 3.3234,
+      "step": 3790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.647033347769598e-05,
+      "loss": 3.2951,
+      "step": 3800
+    },
+    {
+      "epoch": 0.78,
+      "eval_accuracy": 0.3394732251875109,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.5899,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.303,
+      "step": 3800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.603724556084885e-05,
+      "loss": 3.3538,
+      "step": 3810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5604157644001733e-05,
+      "loss": 3.4101,
+      "step": 3820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.517106972715461e-05,
+      "loss": 3.3232,
+      "step": 3830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4737981810307495e-05,
+      "loss": 3.3519,
+      "step": 3840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.430489389346037e-05,
+      "loss": 3.3587,
+      "step": 3850
+    },
+    {
+      "epoch": 0.79,
+      "eval_accuracy": 0.3401709401709402,
+      "eval_loss": 4.19921875,
+      "eval_runtime": 6.5915,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 3850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3871805976613256e-05,
+      "loss": 3.3519,
+      "step": 3860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.3438718059766134e-05,
+      "loss": 3.3635,
+      "step": 3870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.300563014291901e-05,
+      "loss": 3.4013,
+      "step": 3880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2572542226071895e-05,
+      "loss": 3.3224,
+      "step": 3890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.213945430922477e-05,
+      "loss": 3.3237,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8,
+      "eval_accuracy": 0.33935461364032793,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.5894,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.170636639237766e-05,
+      "loss": 3.3706,
+      "step": 3910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1273278475530534e-05,
+      "loss": 3.3024,
+      "step": 3920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.084019055868342e-05,
+      "loss": 3.3717,
+      "step": 3930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.0407102641836295e-05,
+      "loss": 3.35,
+      "step": 3940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.997401472498917e-05,
+      "loss": 3.3136,
+      "step": 3950
+    },
+    {
+      "epoch": 0.81,
+      "eval_accuracy": 0.3393267050409908,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.6033,
+      "eval_samples_per_second": 5.3,
+      "eval_steps_per_second": 0.303,
+      "step": 3950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.954092680814206e-05,
+      "loss": 3.3281,
+      "step": 3960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9107838891294934e-05,
+      "loss": 3.2916,
+      "step": 3970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.867475097444782e-05,
+      "loss": 3.3578,
+      "step": 3980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8241663057600696e-05,
+      "loss": 3.3789,
+      "step": 3990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.780857514075358e-05,
+      "loss": 3.3367,
+      "step": 4000
+    },
+    {
+      "epoch": 0.82,
+      "eval_accuracy": 0.33942438513867085,
+      "eval_loss": 4.203125,
+      "eval_runtime": 6.5852,
+      "eval_samples_per_second": 5.315,
+      "eval_steps_per_second": 0.304,
+      "step": 4000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.737548722390645e-05,
+      "loss": 3.3325,
+      "step": 4010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6942399307059335e-05,
+      "loss": 3.3414,
+      "step": 4020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.650931139021221e-05,
+      "loss": 3.3798,
+      "step": 4030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6076223473365096e-05,
+      "loss": 3.3682,
+      "step": 4040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.564313555651797e-05,
+      "loss": 3.3062,
+      "step": 4050
+    },
+    {
+      "epoch": 0.83,
+      "eval_accuracy": 0.3391034362462934,
+      "eval_loss": 4.1953125,
+      "eval_runtime": 6.5939,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 4050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.521004763967086e-05,
+      "loss": 3.3194,
+      "step": 4060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.4776959722823735e-05,
+      "loss": 3.3226,
+      "step": 4070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.434387180597661e-05,
+      "loss": 3.3148,
+      "step": 4080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3910783889129496e-05,
+      "loss": 3.3504,
+      "step": 4090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3477695972282374e-05,
+      "loss": 3.3112,
+      "step": 4100
+    },
+    {
+      "epoch": 0.84,
+      "eval_accuracy": 0.33969649398220825,
+      "eval_loss": 4.19140625,
+      "eval_runtime": 6.5931,
+      "eval_samples_per_second": 5.309,
+      "eval_steps_per_second": 0.303,
+      "step": 4100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.304460805543526e-05,
+      "loss": 3.3621,
+      "step": 4110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2611520138588135e-05,
+      "loss": 3.3,
+      "step": 4120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.217843222174102e-05,
+      "loss": 3.3523,
+      "step": 4130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.174534430489389e-05,
+      "loss": 3.3182,
+      "step": 4140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1312256388046774e-05,
+      "loss": 3.3345,
+      "step": 4150
+    },
+    {
+      "epoch": 0.85,
+      "eval_accuracy": 0.33911739054596196,
+      "eval_loss": 4.19140625,
+      "eval_runtime": 6.5844,
+      "eval_samples_per_second": 5.316,
+      "eval_steps_per_second": 0.304,
+      "step": 4150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.087916847119965e-05,
+      "loss": 3.2978,
+      "step": 4160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0446080554352535e-05,
+      "loss": 3.2932,
+      "step": 4170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0012992637505416e-05,
+      "loss": 3.3311,
+      "step": 4180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9579904720658297e-05,
+      "loss": 3.3432,
+      "step": 4190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9146816803811177e-05,
+      "loss": 3.3542,
+      "step": 4200
+    },
+    {
+      "epoch": 0.86,
+      "eval_accuracy": 0.33936856793999653,
+      "eval_loss": 4.19140625,
+      "eval_runtime": 6.588,
+      "eval_samples_per_second": 5.313,
+      "eval_steps_per_second": 0.304,
+      "step": 4200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.871372888696405e-05,
+      "loss": 3.341,
+      "step": 4210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8280640970116936e-05,
+      "loss": 3.3403,
+      "step": 4220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7847553053269816e-05,
+      "loss": 3.2958,
+      "step": 4230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7414465136422697e-05,
+      "loss": 3.3407,
+      "step": 4240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6981377219575578e-05,
+      "loss": 3.3262,
+      "step": 4250
+    },
+    {
+      "epoch": 0.87,
+      "eval_accuracy": 0.33954997383568813,
+      "eval_loss": 4.19140625,
+      "eval_runtime": 6.5852,
+      "eval_samples_per_second": 5.315,
+      "eval_steps_per_second": 0.304,
+      "step": 4250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.654828930272846e-05,
+      "loss": 3.2867,
+      "step": 4260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6115201385881332e-05,
+      "loss": 3.3728,
+      "step": 4270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5682113469034213e-05,
+      "loss": 3.3578,
+      "step": 4280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5249025552187094e-05,
+      "loss": 3.3402,
+      "step": 4290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4815937635339975e-05,
+      "loss": 3.3606,
+      "step": 4300
+    },
+    {
+      "epoch": 0.88,
+      "eval_accuracy": 0.33983603697889414,
+      "eval_loss": 4.19140625,
+      "eval_runtime": 6.5958,
+      "eval_samples_per_second": 5.306,
+      "eval_steps_per_second": 0.303,
+      "step": 4300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.4382849718492855e-05,
+      "loss": 3.264,
+      "step": 4310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3949761801645733e-05,
+      "loss": 3.3013,
+      "step": 4320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3516673884798617e-05,
+      "loss": 3.3089,
+      "step": 4330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3083585967951497e-05,
+      "loss": 3.3669,
+      "step": 4340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2650498051104375e-05,
+      "loss": 3.3488,
+      "step": 4350
+    },
+    {
+      "epoch": 0.89,
+      "eval_accuracy": 0.33974533403104834,
+      "eval_loss": 4.19140625,
+      "eval_runtime": 6.6014,
+      "eval_samples_per_second": 5.302,
+      "eval_steps_per_second": 0.303,
+      "step": 4350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.2217410134257256e-05,
+      "loss": 3.334,
+      "step": 4360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1784322217410136e-05,
+      "loss": 3.3529,
+      "step": 4370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1351234300563017e-05,
+      "loss": 3.3454,
+      "step": 4380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0918146383715894e-05,
+      "loss": 3.3431,
+      "step": 4390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0485058466868775e-05,
+      "loss": 3.3803,
+      "step": 4400
+    },
+    {
+      "epoch": 0.9,
+      "eval_accuracy": 0.3396127681841968,
+      "eval_loss": 4.19140625,
+      "eval_runtime": 6.5917,
+      "eval_samples_per_second": 5.31,
+      "eval_steps_per_second": 0.303,
+      "step": 4400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0051970550021656e-05,
+      "loss": 3.3095,
+      "step": 4410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9618882633174533e-05,
+      "loss": 3.3219,
+      "step": 4420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9185794716327414e-05,
+      "loss": 3.3788,
+      "step": 4430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8752706799480295e-05,
+      "loss": 3.3435,
+      "step": 4440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8319618882633175e-05,
+      "loss": 3.3122,
+      "step": 4450
+    },
+    {
+      "epoch": 0.92,
+      "eval_accuracy": 0.33978021978021977,
+      "eval_loss": 4.1875,
+      "eval_runtime": 6.6288,
+      "eval_samples_per_second": 5.28,
+      "eval_steps_per_second": 0.302,
+      "step": 4450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7886530965786056e-05,
+      "loss": 3.2923,
+      "step": 4460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7453443048938937e-05,
+      "loss": 3.3699,
+      "step": 4470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7020355132091814e-05,
+      "loss": 3.2694,
+      "step": 4480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6587267215244695e-05,
+      "loss": 3.2873,
+      "step": 4490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154179298397576e-05,
+      "loss": 3.3429,
+      "step": 4500
+    },
+    {
+      "epoch": 0.93,
+      "eval_accuracy": 0.3399686028257457,
+      "eval_loss": 4.1875,
+      "eval_runtime": 6.6063,
+      "eval_samples_per_second": 5.298,
+      "eval_steps_per_second": 0.303,
+      "step": 4500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5721091381550456e-05,
+      "loss": 3.3116,
+      "step": 4510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5288003464703334e-05,
+      "loss": 3.3643,
+      "step": 4520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4854915547856216e-05,
+      "loss": 3.3477,
+      "step": 4530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4421827631009097e-05,
+      "loss": 3.3393,
+      "step": 4540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3988739714161974e-05,
+      "loss": 3.3114,
+      "step": 4550
+    },
+    {
+      "epoch": 0.94,
+      "eval_accuracy": 0.3400593057735915,
+      "eval_loss": 4.1875,
+      "eval_runtime": 6.5986,
+      "eval_samples_per_second": 5.304,
+      "eval_steps_per_second": 0.303,
+      "step": 4550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3555651797314855e-05,
+      "loss": 3.3285,
+      "step": 4560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3122563880467736e-05,
+      "loss": 3.2873,
+      "step": 4570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2689475963620615e-05,
+      "loss": 3.2914,
+      "step": 4580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2256388046773495e-05,
+      "loss": 3.2841,
+      "step": 4590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1823300129926376e-05,
+      "loss": 3.3,
+      "step": 4600
+    },
+    {
+      "epoch": 0.95,
+      "eval_accuracy": 0.34005232862375717,
+      "eval_loss": 4.1875,
+      "eval_runtime": 6.6023,
+      "eval_samples_per_second": 5.301,
+      "eval_steps_per_second": 0.303,
+      "step": 4600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1390212213079255e-05,
+      "loss": 3.3211,
+      "step": 4610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0957124296232136e-05,
+      "loss": 3.3017,
+      "step": 4620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0524036379385017e-05,
+      "loss": 3.3061,
+      "step": 4630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0090948462537896e-05,
+      "loss": 3.2918,
+      "step": 4640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.657860545690775e-06,
+      "loss": 3.3528,
+      "step": 4650
+    },
+    {
+      "epoch": 0.96,
+      "eval_accuracy": 0.3397941740798884,
+      "eval_loss": 4.1875,
+      "eval_runtime": 6.5894,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 4650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.224772628843655e-06,
+      "loss": 3.3294,
+      "step": 4660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.791684711996536e-06,
+      "loss": 3.3636,
+      "step": 4670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.358596795149417e-06,
+      "loss": 3.371,
+      "step": 4680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.925508878302296e-06,
+      "loss": 3.3523,
+      "step": 4690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.492420961455175e-06,
+      "loss": 3.3195,
+      "step": 4700
+    },
+    {
+      "epoch": 0.97,
+      "eval_accuracy": 0.3398499912785627,
+      "eval_loss": 4.18359375,
+      "eval_runtime": 6.5942,
+      "eval_samples_per_second": 5.308,
+      "eval_steps_per_second": 0.303,
+      "step": 4700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.059333044608056e-06,
+      "loss": 3.3231,
+      "step": 4710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.6262451277609355e-06,
+      "loss": 3.3434,
+      "step": 4720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.193157210913815e-06,
+      "loss": 3.3514,
+      "step": 4730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.803378085751407e-06,
+      "loss": 3.3364,
+      "step": 4740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.370290168904288e-06,
+      "loss": 3.3421,
+      "step": 4750
+    },
+    {
+      "epoch": 0.98,
+      "eval_accuracy": 0.33989185417756845,
+      "eval_loss": 4.18359375,
+      "eval_runtime": 6.6163,
+      "eval_samples_per_second": 5.29,
+      "eval_steps_per_second": 0.302,
+      "step": 4750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.937202252057168e-06,
+      "loss": 3.3033,
+      "step": 4760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.504114335210048e-06,
+      "loss": 3.3042,
+      "step": 4770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.071026418362928e-06,
+      "loss": 3.3248,
+      "step": 4780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6379385015158076e-06,
+      "loss": 3.334,
+      "step": 4790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.204850584668688e-06,
+      "loss": 3.3505,
+      "step": 4800
+    },
+    {
+      "epoch": 0.99,
+      "eval_accuracy": 0.3401500087214373,
+      "eval_loss": 4.18359375,
+      "eval_runtime": 6.5893,
+      "eval_samples_per_second": 5.312,
+      "eval_steps_per_second": 0.304,
+      "step": 4800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.771762667821568e-06,
+      "loss": 3.3231,
+      "step": 4810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.338674750974448e-06,
+      "loss": 3.2872,
+      "step": 4820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9055868341273278e-06,
+      "loss": 3.3428,
+      "step": 4830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.472498917280208e-06,
+      "loss": 3.3162,
+      "step": 4840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.039411000433088e-06,
+      "loss": 3.3606,
+      "step": 4850
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.33996162567591137,
+      "eval_loss": 4.18359375,
+      "eval_runtime": 6.5895,
+      "eval_samples_per_second": 5.311,
+      "eval_steps_per_second": 0.304,
+      "step": 4850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.06323083585968e-07,
+      "loss": 3.3077,
+      "step": 4860
+    },
+    {
+      "epoch": 1.0,
+      "step": 4862,
+      "total_flos": 5.1668865537665925e+19,
+      "train_loss": 3.7094925158997714,
+      "train_runtime": 108243.5499,
+      "train_samples_per_second": 2.875,
+      "train_steps_per_second": 0.045
+    }
+  ],
+  "max_steps": 4862,
+  "num_train_epochs": 1,
+  "total_flos": 5.1668865537665925e+19,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8882a918bcb9a315dfcc77f1a3a879de69b8b68f463ac4b04ba7f0c085598f2d
+size 5115