Training in progress, step 1800

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a9c4b303d98316a33d712a536976501b244eb52e439e922c3dd47139e7e53c12
+oid sha256:4ade53a485d120a2b393a4d31445ac2a0415ccbb10a8b5ea504569156422d103
 size 319977229

checkpoint-1200/adapter_model/adapter_model/README.md

@@ -15,6 +15,17 @@ The following `bitsandbytes` quantization config was used during training:
 - bnb_4bit_use_double_quant: True
 - bnb_4bit_compute_dtype: bfloat16
 
+The following `bitsandbytes` quantization config was used during training:
+- load_in_8bit: False
+- load_in_4bit: True
+- llm_int8_threshold: 6.0
+- llm_int8_skip_modules: None
+- llm_int8_enable_fp32_cpu_offload: False
+- llm_int8_has_fp16_weight: False
+- bnb_4bit_quant_type: nf4
+- bnb_4bit_use_double_quant: True
+- bnb_4bit_compute_dtype: bfloat16
+
 The following `bitsandbytes` quantization config was used during training:
 - load_in_8bit: False
 - load_in_4bit: True
@@ -27,6 +38,7 @@ The following `bitsandbytes` quantization config was used during training:
 - bnb_4bit_compute_dtype: bfloat16
 ### Framework versions
 
+- PEFT 0.4.0
 - PEFT 0.4.0
 
 - PEFT 0.4.0

checkpoint-1200/adapter_model/adapter_model/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:395a9b72b3abe980766956d1ba207fa42c62ef7dbd8aa756f04a7162b2781178
+oid sha256:a9c4b303d98316a33d712a536976501b244eb52e439e922c3dd47139e7e53c12
 size 319977229

checkpoint-1800/README.md

@@ -0,0 +1,20 @@
+---
+library_name: peft
+---
+## Training procedure
+
+
+The following `bitsandbytes` quantization config was used during training:
+- load_in_8bit: False
+- load_in_4bit: True
+- llm_int8_threshold: 6.0
+- llm_int8_skip_modules: None
+- llm_int8_enable_fp32_cpu_offload: False
+- llm_int8_has_fp16_weight: False
+- bnb_4bit_quant_type: nf4
+- bnb_4bit_use_double_quant: True
+- bnb_4bit_compute_dtype: bfloat16
+### Framework versions
+
+
+- PEFT 0.4.0

checkpoint-1800/adapter_config.json

@@ -0,0 +1,26 @@
+{
+ "auto_mapping": null,
+ "base_model_name_or_path": "NousResearch/Nous-Hermes-llama-2-7b",
+ "bias": "none",
+ "fan_in_fan_out": false,
+ "inference_mode": true,
+ "init_lora_weights": true,
+ "layers_pattern": null,
+ "layers_to_transform": null,
+ "lora_alpha": 16.0,
+ "lora_dropout": 0.1,
+ "modules_to_save": null,
+ "peft_type": "LORA",
+ "r": 64,
+ "revision": null,
+ "target_modules": [
+ "q_proj",
+ "v_proj",
+ "o_proj",
+ "gate_proj",
+ "k_proj",
+ "up_proj",
+ "down_proj"
+ ],
+ "task_type": "CAUSAL_LM"
+}

checkpoint-1800/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ade53a485d120a2b393a4d31445ac2a0415ccbb10a8b5ea504569156422d103
+size 319977229

checkpoint-1800/added_tokens.json

@@ -0,0 +1,3 @@
+{
+ "<pad>": 32000
+}

checkpoint-1800/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9083b2dea8016cc77066d3e9750b4ef8139dcff3fb5d10ad62f3f2c28c28961b
+size 1279539973

checkpoint-1800/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:338596f81d36de67f0dc44fc9ca3bf511bc66a295beacae1e7207d950e154bea
+size 14511

checkpoint-1800/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0e0a1736a46fd1627af3c246e44261aaac909256abbd413b5ee5c968f6b2d8e
+size 627

checkpoint-1800/special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+ "bos_token": "<s>",
+ "eos_token": "</s>",
+ "pad_token": "<unk>",
+ "unk_token": "<unk>"
+}

checkpoint-1800/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

checkpoint-1800/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
+ },
+ "legacy": false,
+ "model_max_length": 1000000000000000019884624838656,
+ "pad_token": null,
+ "padding_side": "right",
+ "sp_model_kwargs": {},
+ "tokenizer_class": "LlamaTokenizer",
+ "unk_token": {
+ "__type": "AddedToken",
+ "content": "<unk>",
+ "lstrip": false,
+ "normalized": true,
+ "rstrip": false,
+ "single_word": false
+ }
+}

checkpoint-1800/trainer_state.json

@@ -0,0 +1,1735 @@
+{
+ "best_metric": 0.5305746793746948,
+ "best_model_checkpoint": "./output_v2/7b_cluster024_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_024/checkpoint-1200",
+ "epoch": 2.9850746268656714,
+ "global_step": 1800,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.02,
+ "learning_rate": 0.0002,
+ "loss": 0.6328,
+ "step": 10
+ },
+ {
+ "epoch": 0.03,
+ "learning_rate": 0.0002,
+ "loss": 0.616,
+ "step": 20
+ },
+ {
+ "epoch": 0.05,
+ "learning_rate": 0.0002,
+ "loss": 0.6065,
+ "step": 30
+ },
+ {
+ "epoch": 0.07,
+ "learning_rate": 0.0002,
+ "loss": 0.5889,
+ "step": 40
+ },
+ {
+ "epoch": 0.08,
+ "learning_rate": 0.0002,
+ "loss": 0.6014,
+ "step": 50
+ },
+ {
+ "epoch": 0.1,
+ "learning_rate": 0.0002,
+ "loss": 0.5784,
+ "step": 60
+ },
+ {
+ "epoch": 0.12,
+ "learning_rate": 0.0002,
+ "loss": 0.5916,
+ "step": 70
+ },
+ {
+ "epoch": 0.13,
+ "learning_rate": 0.0002,
+ "loss": 0.5858,
+ "step": 80
+ },
+ {
+ "epoch": 0.15,
+ "learning_rate": 0.0002,
+ "loss": 0.5908,
+ "step": 90
+ },
+ {
+ "epoch": 0.17,
+ "learning_rate": 0.0002,
+ "loss": 0.5695,
+ "step": 100
+ },
+ {
+ "epoch": 0.18,
+ "learning_rate": 0.0002,
+ "loss": 0.5748,
+ "step": 110
+ },
+ {
+ "epoch": 0.2,
+ "learning_rate": 0.0002,
+ "loss": 0.5461,
+ "step": 120
+ },
+ {
+ "epoch": 0.22,
+ "learning_rate": 0.0002,
+ "loss": 0.5738,
+ "step": 130
+ },
+ {
+ "epoch": 0.23,
+ "learning_rate": 0.0002,
+ "loss": 0.544,
+ "step": 140
+ },
+ {
+ "epoch": 0.25,
+ "learning_rate": 0.0002,
+ "loss": 0.5646,
+ "step": 150
+ },
+ {
+ "epoch": 0.27,
+ "learning_rate": 0.0002,
+ "loss": 0.5767,
+ "step": 160
+ },
+ {
+ "epoch": 0.28,
+ "learning_rate": 0.0002,
+ "loss": 0.5783,
+ "step": 170
+ },
+ {
+ "epoch": 0.3,
+ "learning_rate": 0.0002,
+ "loss": 0.5553,
+ "step": 180
+ },
+ {
+ "epoch": 0.32,
+ "learning_rate": 0.0002,
+ "loss": 0.5455,
+ "step": 190
+ },
+ {
+ "epoch": 0.33,
+ "learning_rate": 0.0002,
+ "loss": 0.5544,
+ "step": 200
+ },
+ {
+ "epoch": 0.33,
+ "eval_loss": 0.5742304921150208,
+ "eval_runtime": 241.4438,
+ "eval_samples_per_second": 4.142,
+ "eval_steps_per_second": 2.071,
+ "step": 200
+ },
+ {
+ "epoch": 0.33,
+ "mmlu_eval_accuracy": 0.4702447044631963,
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+ "mmlu_eval_accuracy_astronomy": 0.375,
+ "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
+ "mmlu_eval_accuracy_college_biology": 0.5,
+ "mmlu_eval_accuracy_college_chemistry": 0.0,
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
+ "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+ "mmlu_eval_accuracy_electrical_engineering": 0.375,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
+ "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.34375,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+ "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
+ "mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
+ "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
+ "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+ "mmlu_eval_accuracy_international_law": 0.6923076923076923,
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+ "mmlu_eval_accuracy_management": 0.5454545454545454,
+ "mmlu_eval_accuracy_marketing": 0.72,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
+ "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
+ "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
+ "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
+ "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
+ "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
+ "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+ "mmlu_eval_accuracy_security_studies": 0.5555555555555556,
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.171094535030522,
+ "step": 200
+ },
+ {
+ "epoch": 0.35,
+ "learning_rate": 0.0002,
+ "loss": 0.5533,
+ "step": 210
+ },
+ {
+ "epoch": 0.36,
+ "learning_rate": 0.0002,
+ "loss": 0.5512,
+ "step": 220
+ },
+ {
+ "epoch": 0.38,
+ "learning_rate": 0.0002,
+ "loss": 0.5598,
+ "step": 230
+ },
+ {
+ "epoch": 0.4,
+ "learning_rate": 0.0002,
+ "loss": 0.5616,
+ "step": 240
+ },
+ {
+ "epoch": 0.41,
+ "learning_rate": 0.0002,
+ "loss": 0.5501,
+ "step": 250
+ },
+ {
+ "epoch": 0.43,
+ "learning_rate": 0.0002,
+ "loss": 0.5391,
+ "step": 260
+ },
+ {
+ "epoch": 0.45,
+ "learning_rate": 0.0002,
+ "loss": 0.5608,
+ "step": 270
+ },
+ {
+ "epoch": 0.46,
+ "learning_rate": 0.0002,
+ "loss": 0.5412,
+ "step": 280
+ },
+ {
+ "epoch": 0.48,
+ "learning_rate": 0.0002,
+ "loss": 0.5582,
+ "step": 290
+ },
+ {
+ "epoch": 0.5,
+ "learning_rate": 0.0002,
+ "loss": 0.5479,
+ "step": 300
+ },
+ {
+ "epoch": 0.51,
+ "learning_rate": 0.0002,
+ "loss": 0.552,
+ "step": 310
+ },
+ {
+ "epoch": 0.53,
+ "learning_rate": 0.0002,
+ "loss": 0.5618,
+ "step": 320
+ },
+ {
+ "epoch": 0.55,
+ "learning_rate": 0.0002,
+ "loss": 0.5535,
+ "step": 330
+ },
+ {
+ "epoch": 0.56,
+ "learning_rate": 0.0002,
+ "loss": 0.5523,
+ "step": 340
+ },
+ {
+ "epoch": 0.58,
+ "learning_rate": 0.0002,
+ "loss": 0.5497,
+ "step": 350
+ },
+ {
+ "epoch": 0.6,
+ "learning_rate": 0.0002,
+ "loss": 0.5473,
+ "step": 360
+ },
+ {
+ "epoch": 0.61,
+ "learning_rate": 0.0002,
+ "loss": 0.5481,
+ "step": 370
+ },
+ {
+ "epoch": 0.63,
+ "learning_rate": 0.0002,
+ "loss": 0.537,
+ "step": 380
+ },
+ {
+ "epoch": 0.65,
+ "learning_rate": 0.0002,
+ "loss": 0.5456,
+ "step": 390
+ },
+ {
+ "epoch": 0.66,
+ "learning_rate": 0.0002,
+ "loss": 0.5572,
+ "step": 400
+ },
+ {
+ "epoch": 0.66,
+ "eval_loss": 0.5586665868759155,
+ "eval_runtime": 241.3904,
+ "eval_samples_per_second": 4.143,
+ "eval_steps_per_second": 2.071,
+ "step": 400
+ },
+ {
+ "epoch": 0.66,
+ "mmlu_eval_accuracy": 0.46203622406572054,
+ "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
+ "mmlu_eval_accuracy_astronomy": 0.375,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
+ "mmlu_eval_accuracy_college_biology": 0.4375,
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
+ "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
+ "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+ "mmlu_eval_accuracy_electrical_engineering": 0.3125,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
+ "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.375,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
+ "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
+ "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
+ "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+ "mmlu_eval_accuracy_international_law": 0.7692307692307693,
+ "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+ "mmlu_eval_accuracy_management": 0.45454545454545453,
+ "mmlu_eval_accuracy_marketing": 0.72,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
+ "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+ "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
+ "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
+ "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
+ "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
+ "mmlu_eval_accuracy_sociology": 0.6363636363636364,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.1594297412792633,
+ "step": 400
+ },
+ {
+ "epoch": 0.68,
+ "learning_rate": 0.0002,
+ "loss": 0.5303,
+ "step": 410
+ },
+ {
+ "epoch": 0.7,
+ "learning_rate": 0.0002,
+ "loss": 0.5266,
+ "step": 420
+ },
+ {
+ "epoch": 0.71,
+ "learning_rate": 0.0002,
+ "loss": 0.5356,
+ "step": 430
+ },
+ {
+ "epoch": 0.73,
+ "learning_rate": 0.0002,
+ "loss": 0.5238,
+ "step": 440
+ },
+ {
+ "epoch": 0.75,
+ "learning_rate": 0.0002,
+ "loss": 0.5483,
+ "step": 450
+ },
+ {
+ "epoch": 0.76,
+ "learning_rate": 0.0002,
+ "loss": 0.5582,
+ "step": 460
+ },
+ {
+ "epoch": 0.78,
+ "learning_rate": 0.0002,
+ "loss": 0.5259,
+ "step": 470
+ },
+ {
+ "epoch": 0.8,
+ "learning_rate": 0.0002,
+ "loss": 0.5422,
+ "step": 480
+ },
+ {
+ "epoch": 0.81,
+ "learning_rate": 0.0002,
+ "loss": 0.5357,
+ "step": 490
+ },
+ {
+ "epoch": 0.83,
+ "learning_rate": 0.0002,
+ "loss": 0.5436,
+ "step": 500
+ },
+ {
+ "epoch": 0.85,
+ "learning_rate": 0.0002,
+ "loss": 0.5274,
+ "step": 510
+ },
+ {
+ "epoch": 0.86,
+ "learning_rate": 0.0002,
+ "loss": 0.5355,
+ "step": 520
+ },
+ {
+ "epoch": 0.88,
+ "learning_rate": 0.0002,
+ "loss": 0.542,
+ "step": 530
+ },
+ {
+ "epoch": 0.9,
+ "learning_rate": 0.0002,
+ "loss": 0.5333,
+ "step": 540
+ },
+ {
+ "epoch": 0.91,
+ "learning_rate": 0.0002,
+ "loss": 0.5327,
+ "step": 550
+ },
+ {
+ "epoch": 0.93,
+ "learning_rate": 0.0002,
+ "loss": 0.554,
+ "step": 560
+ },
+ {
+ "epoch": 0.95,
+ "learning_rate": 0.0002,
+ "loss": 0.5313,
+ "step": 570
+ },
+ {
+ "epoch": 0.96,
+ "learning_rate": 0.0002,
+ "loss": 0.5244,
+ "step": 580
+ },
+ {
+ "epoch": 0.98,
+ "learning_rate": 0.0002,
+ "loss": 0.5202,
+ "step": 590
+ },
+ {
+ "epoch": 1.0,
+ "learning_rate": 0.0002,
+ "loss": 0.5363,
+ "step": 600
+ },
+ {
+ "epoch": 1.0,
+ "eval_loss": 0.5475168228149414,
+ "eval_runtime": 241.4806,
+ "eval_samples_per_second": 4.141,
+ "eval_steps_per_second": 2.071,
+ "step": 600
+ },
+ {
+ "epoch": 1.0,
+ "mmlu_eval_accuracy": 0.47075667666126014,
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+ "mmlu_eval_accuracy_astronomy": 0.3125,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
+ "mmlu_eval_accuracy_college_biology": 0.4375,
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
+ "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
+ "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.375,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
+ "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
+ "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
+ "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
+ "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+ "mmlu_eval_accuracy_international_law": 0.6923076923076923,
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+ "mmlu_eval_accuracy_management": 0.6363636363636364,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
+ "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
+ "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
+ "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
+ "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
+ "mmlu_eval_accuracy_virology": 0.3888888888888889,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.2061605295920186,
+ "step": 600
+ },
+ {
+ "epoch": 1.01,
+ "learning_rate": 0.0002,
+ "loss": 0.4962,
+ "step": 610
+ },
+ {
+ "epoch": 1.03,
+ "learning_rate": 0.0002,
+ "loss": 0.4933,
+ "step": 620
+ },
+ {
+ "epoch": 1.04,
+ "learning_rate": 0.0002,
+ "loss": 0.4587,
+ "step": 630
+ },
+ {
+ "epoch": 1.06,
+ "learning_rate": 0.0002,
+ "loss": 0.4631,
+ "step": 640
+ },
+ {
+ "epoch": 1.08,
+ "learning_rate": 0.0002,
+ "loss": 0.4595,
+ "step": 650
+ },
+ {
+ "epoch": 1.09,
+ "learning_rate": 0.0002,
+ "loss": 0.4698,
+ "step": 660
+ },
+ {
+ "epoch": 1.11,
+ "learning_rate": 0.0002,
+ "loss": 0.4726,
+ "step": 670
+ },
+ {
+ "epoch": 1.13,
+ "learning_rate": 0.0002,
+ "loss": 0.4887,
+ "step": 680
+ },
+ {
+ "epoch": 1.14,
+ "learning_rate": 0.0002,
+ "loss": 0.4819,
+ "step": 690
+ },
+ {
+ "epoch": 1.16,
+ "learning_rate": 0.0002,
+ "loss": 0.4638,
+ "step": 700
+ },
+ {
+ "epoch": 1.18,
+ "learning_rate": 0.0002,
+ "loss": 0.4656,
+ "step": 710
+ },
+ {
+ "epoch": 1.19,
+ "learning_rate": 0.0002,
+ "loss": 0.4932,
+ "step": 720
+ },
+ {
+ "epoch": 1.21,
+ "learning_rate": 0.0002,
+ "loss": 0.4762,
+ "step": 730
+ },
+ {
+ "epoch": 1.23,
+ "learning_rate": 0.0002,
+ "loss": 0.4805,
+ "step": 740
+ },
+ {
+ "epoch": 1.24,
+ "learning_rate": 0.0002,
+ "loss": 0.4746,
+ "step": 750
+ },
+ {
+ "epoch": 1.26,
+ "learning_rate": 0.0002,
+ "loss": 0.4883,
+ "step": 760
+ },
+ {
+ "epoch": 1.28,
+ "learning_rate": 0.0002,
+ "loss": 0.4714,
+ "step": 770
+ },
+ {
+ "epoch": 1.29,
+ "learning_rate": 0.0002,
+ "loss": 0.4664,
+ "step": 780
+ },
+ {
+ "epoch": 1.31,
+ "learning_rate": 0.0002,
+ "loss": 0.4667,
+ "step": 790
+ },
+ {
+ "epoch": 1.33,
+ "learning_rate": 0.0002,
+ "loss": 0.477,
+ "step": 800
+ },
+ {
+ "epoch": 1.33,
+ "eval_loss": 0.5436688661575317,
+ "eval_runtime": 241.2705,
+ "eval_samples_per_second": 4.145,
+ "eval_steps_per_second": 2.072,
+ "step": 800
+ },
+ {
+ "epoch": 1.33,
+ "mmlu_eval_accuracy": 0.4770226329405574,
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
+ "mmlu_eval_accuracy_astronomy": 0.375,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
+ "mmlu_eval_accuracy_college_biology": 0.4375,
+ "mmlu_eval_accuracy_college_chemistry": 0.25,
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
+ "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
+ "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.375,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
+ "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
+ "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
+ "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
+ "mmlu_eval_accuracy_international_law": 0.7692307692307693,
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
+ "mmlu_eval_accuracy_management": 0.5454545454545454,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
+ "mmlu_eval_accuracy_moral_scenarios": 0.23,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
+ "mmlu_eval_accuracy_professional_accounting": 0.3225806451612903,
+ "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
+ "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
+ "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+ "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.2083673925763945,
+ "step": 800
+ },
+ {
+ "epoch": 1.34,
+ "learning_rate": 0.0002,
+ "loss": 0.4773,
+ "step": 810
+ },
+ {
+ "epoch": 1.36,
+ "learning_rate": 0.0002,
+ "loss": 0.4715,
+ "step": 820
+ },
+ {
+ "epoch": 1.38,
+ "learning_rate": 0.0002,
+ "loss": 0.4684,
+ "step": 830
+ },
+ {
+ "epoch": 1.39,
+ "learning_rate": 0.0002,
+ "loss": 0.4794,
+ "step": 840
+ },
+ {
+ "epoch": 1.41,
+ "learning_rate": 0.0002,
+ "loss": 0.4626,
+ "step": 850
+ },
+ {
+ "epoch": 1.43,
+ "learning_rate": 0.0002,
+ "loss": 0.4697,
+ "step": 860
+ },
+ {
+ "epoch": 1.44,
+ "learning_rate": 0.0002,
+ "loss": 0.4851,
+ "step": 870
+ },
+ {
+ "epoch": 1.46,
+ "learning_rate": 0.0002,
+ "loss": 0.4641,
+ "step": 880
+ },
+ {
+ "epoch": 1.48,
+ "learning_rate": 0.0002,
+ "loss": 0.4744,
+ "step": 890
+ },
+ {
+ "epoch": 1.49,
+ "learning_rate": 0.0002,
+ "loss": 0.4579,
+ "step": 900
+ },
+ {
+ "epoch": 1.51,
+ "learning_rate": 0.0002,
+ "loss": 0.4723,
+ "step": 910
+ },
+ {
+ "epoch": 1.53,
+ "learning_rate": 0.0002,
+ "loss": 0.471,
+ "step": 920
+ },
+ {
+ "epoch": 1.54,
+ "learning_rate": 0.0002,
+ "loss": 0.4549,
+ "step": 930
+ },
+ {
+ "epoch": 1.56,
+ "learning_rate": 0.0002,
+ "loss": 0.4704,
+ "step": 940
+ },
+ {
+ "epoch": 1.58,
+ "learning_rate": 0.0002,
+ "loss": 0.4697,
+ "step": 950
+ },
+ {
+ "epoch": 1.59,
+ "learning_rate": 0.0002,
+ "loss": 0.4725,
+ "step": 960
+ },
+ {
+ "epoch": 1.61,
+ "learning_rate": 0.0002,
+ "loss": 0.4727,
+ "step": 970
+ },
+ {
+ "epoch": 1.63,
+ "learning_rate": 0.0002,
+ "loss": 0.4707,
+ "step": 980
+ },
+ {
+ "epoch": 1.64,
+ "learning_rate": 0.0002,
+ "loss": 0.4727,
+ "step": 990
+ },
+ {
+ "epoch": 1.66,
+ "learning_rate": 0.0002,
+ "loss": 0.488,
+ "step": 1000
+ },
+ {
+ "epoch": 1.66,
+ "eval_loss": 0.5379906296730042,
+ "eval_runtime": 241.646,
+ "eval_samples_per_second": 4.138,
+ "eval_steps_per_second": 2.069,
+ "step": 1000
+ },
+ {
+ "epoch": 1.66,
+ "mmlu_eval_accuracy": 0.4825734797626005,
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
+ "mmlu_eval_accuracy_astronomy": 0.3125,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
+ "mmlu_eval_accuracy_college_biology": 0.375,
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+ "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
+ "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+ "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
+ "mmlu_eval_accuracy_electrical_engineering": 0.5,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.3170731707317073,
+ "mmlu_eval_accuracy_formal_logic": 0.42857142857142855,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.4375,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5,
+ "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
+ "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
+ "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+ "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+ "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+ "mmlu_eval_accuracy_management": 0.6363636363636364,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
+ "mmlu_eval_accuracy_moral_scenarios": 0.28,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
+ "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
+ "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
+ "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
+ "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+ "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
+ "mmlu_eval_accuracy_virology": 0.5,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.3014565500827122,
+ "step": 1000
+ },
+ {
+ "epoch": 1.67,
+ "learning_rate": 0.0002,
+ "loss": 0.457,
+ "step": 1010
+ },
+ {
+ "epoch": 1.69,
+ "learning_rate": 0.0002,
+ "loss": 0.4595,
+ "step": 1020
+ },
+ {
+ "epoch": 1.71,
+ "learning_rate": 0.0002,
+ "loss": 0.4713,
+ "step": 1030
+ },
+ {
+ "epoch": 1.72,
+ "learning_rate": 0.0002,
+ "loss": 0.4642,
+ "step": 1040
+ },
+ {
+ "epoch": 1.74,
+ "learning_rate": 0.0002,
+ "loss": 0.4771,
+ "step": 1050
+ },
+ {
+ "epoch": 1.76,
+ "learning_rate": 0.0002,
+ "loss": 0.4773,
+ "step": 1060
+ },
+ {
+ "epoch": 1.77,
+ "learning_rate": 0.0002,
+ "loss": 0.4584,
+ "step": 1070
+ },
+ {
+ "epoch": 1.79,
+ "learning_rate": 0.0002,
+ "loss": 0.4663,
+ "step": 1080
+ },
+ {
+ "epoch": 1.81,
+ "learning_rate": 0.0002,
+ "loss": 0.4566,
+ "step": 1090
+ },
+ {
+ "epoch": 1.82,
+ "learning_rate": 0.0002,
+ "loss": 0.4671,
+ "step": 1100
+ },
+ {
+ "epoch": 1.84,
+ "learning_rate": 0.0002,
+ "loss": 0.4562,
+ "step": 1110
+ },
+ {
+ "epoch": 1.86,
+ "learning_rate": 0.0002,
+ "loss": 0.4607,
+ "step": 1120
+ },
+ {
+ "epoch": 1.87,
+ "learning_rate": 0.0002,
+ "loss": 0.4764,
+ "step": 1130
+ },
+ {
+ "epoch": 1.89,
+ "learning_rate": 0.0002,
+ "loss": 0.4594,
+ "step": 1140
+ },
+ {
+ "epoch": 1.91,
+ "learning_rate": 0.0002,
+ "loss": 0.4579,
+ "step": 1150
+ },
+ {
+ "epoch": 1.92,
+ "learning_rate": 0.0002,
+ "loss": 0.4536,
+ "step": 1160
+ },
+ {
+ "epoch": 1.94,
+ "learning_rate": 0.0002,
+ "loss": 0.4582,
+ "step": 1170
+ },
+ {
+ "epoch": 1.96,
+ "learning_rate": 0.0002,
+ "loss": 0.4629,
+ "step": 1180
+ },
+ {
+ "epoch": 1.97,
+ "learning_rate": 0.0002,
+ "loss": 0.4682,
+ "step": 1190
+ },
+ {
+ "epoch": 1.99,
+ "learning_rate": 0.0002,
+ "loss": 0.4491,
+ "step": 1200
+ },
+ {
+ "epoch": 1.99,
+ "eval_loss": 0.5305746793746948,
+ "eval_runtime": 241.2768,
+ "eval_samples_per_second": 4.145,
+ "eval_steps_per_second": 2.072,
+ "step": 1200
+ },
+ {
+ "epoch": 1.99,
+ "mmlu_eval_accuracy": 0.47944607769691217,
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
+ "mmlu_eval_accuracy_astronomy": 0.3125,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
+ "mmlu_eval_accuracy_college_biology": 0.4375,
+ "mmlu_eval_accuracy_college_chemistry": 0.25,
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+ "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
+ "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+ "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
+ "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.40625,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.2727272727272727,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.3076923076923077,
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
+ "mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
+ "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+ "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+ "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
+ "mmlu_eval_accuracy_moral_scenarios": 0.27,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
+ "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
+ "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
+ "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
+ "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
+ "mmlu_eval_accuracy_public_relations": 0.5,
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
+ "mmlu_eval_accuracy_virology": 0.3333333333333333,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.2915717554388095,
+ "step": 1200
+ },
+ {
+ "epoch": 2.01,
+ "learning_rate": 0.0002,
+ "loss": 0.4285,
+ "step": 1210
+ },
+ {
+ "epoch": 2.02,
+ "learning_rate": 0.0002,
+ "loss": 0.3863,
+ "step": 1220
+ },
+ {
+ "epoch": 2.04,
+ "learning_rate": 0.0002,
+ "loss": 0.374,
+ "step": 1230
+ },
+ {
+ "epoch": 2.06,
+ "learning_rate": 0.0002,
+ "loss": 0.3699,
+ "step": 1240
+ },
+ {
+ "epoch": 2.07,
+ "learning_rate": 0.0002,
+ "loss": 0.3627,
+ "step": 1250
+ },
+ {
+ "epoch": 2.09,
+ "learning_rate": 0.0002,
+ "loss": 0.3736,
+ "step": 1260
+ },
+ {
+ "epoch": 2.11,
+ "learning_rate": 0.0002,
+ "loss": 0.376,
+ "step": 1270
+ },
+ {
+ "epoch": 2.12,
+ "learning_rate": 0.0002,
+ "loss": 0.3631,
+ "step": 1280
+ },
+ {
+ "epoch": 2.14,
+ "learning_rate": 0.0002,
+ "loss": 0.3657,
+ "step": 1290
+ },
+ {
+ "epoch": 2.16,
+ "learning_rate": 0.0002,
+ "loss": 0.3773,
+ "step": 1300
+ },
+ {
+ "epoch": 2.17,
+ "learning_rate": 0.0002,
+ "loss": 0.3859,
+ "step": 1310
+ },
+ {
+ "epoch": 2.19,
+ "learning_rate": 0.0002,
+ "loss": 0.4034,
+ "step": 1320
+ },
+ {
+ "epoch": 2.21,
+ "learning_rate": 0.0002,
+ "loss": 0.3904,
+ "step": 1330
+ },
+ {
+ "epoch": 2.22,
+ "learning_rate": 0.0002,
+ "loss": 0.3787,
+ "step": 1340
+ },
+ {
+ "epoch": 2.24,
+ "learning_rate": 0.0002,
+ "loss": 0.3739,
+ "step": 1350
+ },
+ {
+ "epoch": 2.26,
+ "learning_rate": 0.0002,
+ "loss": 0.3616,
+ "step": 1360
+ },
+ {
+ "epoch": 2.27,
+ "learning_rate": 0.0002,
+ "loss": 0.3867,
+ "step": 1370
+ },
+ {
+ "epoch": 2.29,
+ "learning_rate": 0.0002,
+ "loss": 0.3666,
+ "step": 1380
+ },
+ {
+ "epoch": 2.31,
+ "learning_rate": 0.0002,
+ "loss": 0.3752,
+ "step": 1390
+ },
+ {
+ "epoch": 2.32,
+ "learning_rate": 0.0002,
+ "loss": 0.3864,
+ "step": 1400
+ },
+ {
+ "epoch": 2.32,
+ "eval_loss": 0.5526766180992126,
+ "eval_runtime": 241.8332,
+ "eval_samples_per_second": 4.135,
+ "eval_steps_per_second": 2.068,
+ "step": 1400
+ },
+ {
+ "epoch": 2.32,
+ "mmlu_eval_accuracy": 0.489128246112734,
+ "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
+ "mmlu_eval_accuracy_astronomy": 0.375,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
+ "mmlu_eval_accuracy_college_biology": 0.375,
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+ "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
+ "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+ "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
+ "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
+ "mmlu_eval_accuracy_formal_logic": 0.5,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.40625,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
+ "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
+ "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+ "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
+ "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
+ "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6395348837209303,
+ "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
+ "mmlu_eval_accuracy_moral_scenarios": 0.28,
+ "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
+ "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
+ "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
+ "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
+ "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
+ "mmlu_eval_accuracy_public_relations": 0.5,
+ "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
+ "mmlu_eval_accuracy_sociology": 0.7272727272727273,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
+ "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
+ "mmlu_loss": 1.2334477945812066,
+ "step": 1400
+ },
+ {
+ "epoch": 2.34,
+ "learning_rate": 0.0002,
+ "loss": 0.3752,
+ "step": 1410
+ },
+ {
+ "epoch": 2.35,
+ "learning_rate": 0.0002,
+ "loss": 0.3874,
+ "step": 1420
+ },
+ {
+ "epoch": 2.37,
+ "learning_rate": 0.0002,
+ "loss": 0.3832,
+ "step": 1430
+ },
+ {
+ "epoch": 2.39,
+ "learning_rate": 0.0002,
+ "loss": 0.3793,
+ "step": 1440
+ },
+ {
+ "epoch": 2.4,
+ "learning_rate": 0.0002,
+ "loss": 0.3768,
+ "step": 1450
+ },
+ {
+ "epoch": 2.42,
+ "learning_rate": 0.0002,
+ "loss": 0.3692,
+ "step": 1460
+ },
+ {
+ "epoch": 2.44,
+ "learning_rate": 0.0002,
+ "loss": 0.3792,
+ "step": 1470
+ },
+ {
+ "epoch": 2.45,
+ "learning_rate": 0.0002,
+ "loss": 0.3724,
+ "step": 1480
+ },
+ {
+ "epoch": 2.47,
+ "learning_rate": 0.0002,
+ "loss": 0.3905,
+ "step": 1490
+ },
+ {
+ "epoch": 2.49,
+ "learning_rate": 0.0002,
+ "loss": 0.3848,
+ "step": 1500
+ },
+ {
+ "epoch": 2.5,
+ "learning_rate": 0.0002,
+ "loss": 0.3701,
+ "step": 1510
+ },
+ {
+ "epoch": 2.52,
+ "learning_rate": 0.0002,
+ "loss": 0.3838,
+ "step": 1520
+ },
+ {
+ "epoch": 2.54,
+ "learning_rate": 0.0002,
+ "loss": 0.3833,
+ "step": 1530
+ },
+ {
+ "epoch": 2.55,
+ "learning_rate": 0.0002,
+ "loss": 0.3947,
+ "step": 1540
+ },
+ {
+ "epoch": 2.57,
+ "learning_rate": 0.0002,
+ "loss": 0.3625,
+ "step": 1550
+ },
+ {
+ "epoch": 2.59,
+ "learning_rate": 0.0002,
+ "loss": 0.3724,
+ "step": 1560
+ },
+ {
+ "epoch": 2.6,
+ "learning_rate": 0.0002,
+ "loss": 0.375,
+ "step": 1570
+ },
+ {
+ "epoch": 2.62,
+ "learning_rate": 0.0002,
+ "loss": 0.3758,
+ "step": 1580
+ },
+ {
+ "epoch": 2.64,
+ "learning_rate": 0.0002,
+ "loss": 0.3917,
+ "step": 1590
+ },
+ {
+ "epoch": 2.65,
+ "learning_rate": 0.0002,
+ "loss": 0.3782,
+ "step": 1600
+ },
+ {
+ "epoch": 2.65,
+ "eval_loss": 0.5458189845085144,
+ "eval_runtime": 241.2525,
+ "eval_samples_per_second": 4.145,
+ "eval_steps_per_second": 2.073,
+ "step": 1600
+ },
+ {
+ "epoch": 2.65,
+ "mmlu_eval_accuracy": 0.48696867818397405,
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
+ "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
+ "mmlu_eval_accuracy_astronomy": 0.4375,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
+ "mmlu_eval_accuracy_college_biology": 0.375,
+ "mmlu_eval_accuracy_college_chemistry": 0.25,
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
+ "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
+ "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
+ "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
+ "mmlu_eval_accuracy_formal_logic": 0.5,
+ "mmlu_eval_accuracy_global_facts": 0.4,
+ "mmlu_eval_accuracy_high_school_biology": 0.40625,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5,
+ "mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3953488372093023,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
+ "mmlu_eval_accuracy_high_school_psychology": 0.75,
+ "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
+ "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
+ "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
+ "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+ "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+ "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+ "mmlu_eval_accuracy_management": 0.6363636363636364,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
+ "mmlu_eval_accuracy_moral_scenarios": 0.29,
+ "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
+ "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
+ "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
+ "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
+ "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
+ "mmlu_eval_accuracy_public_relations": 0.5,
+ "mmlu_eval_accuracy_security_studies": 0.4074074074074074,
+ "mmlu_eval_accuracy_sociology": 0.7272727272727273,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
+ "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
+ "mmlu_loss": 1.3606003118440626,
+ "step": 1600
+ },
+ {
+ "epoch": 2.67,
+ "learning_rate": 0.0002,
+ "loss": 0.387,
+ "step": 1610
+ },
+ {
+ "epoch": 2.69,
+ "learning_rate": 0.0002,
+ "loss": 0.3856,
+ "step": 1620
+ },
+ {
+ "epoch": 2.7,
+ "learning_rate": 0.0002,
+ "loss": 0.3713,
+ "step": 1630
+ },
+ {
+ "epoch": 2.72,
+ "learning_rate": 0.0002,
+ "loss": 0.3666,
+ "step": 1640
+ },
+ {
+ "epoch": 2.74,
+ "learning_rate": 0.0002,
+ "loss": 0.379,
+ "step": 1650
+ },
+ {
+ "epoch": 2.75,
+ "learning_rate": 0.0002,
+ "loss": 0.3646,
+ "step": 1660
+ },
+ {
+ "epoch": 2.77,
+ "learning_rate": 0.0002,
+ "loss": 0.3771,
+ "step": 1670
+ },
+ {
+ "epoch": 2.79,
+ "learning_rate": 0.0002,
+ "loss": 0.3751,
+ "step": 1680
+ },
+ {
+ "epoch": 2.8,
+ "learning_rate": 0.0002,
+ "loss": 0.3757,
+ "step": 1690
+ },
+ {
+ "epoch": 2.82,
+ "learning_rate": 0.0002,
+ "loss": 0.3796,
+ "step": 1700
+ },
+ {
+ "epoch": 2.84,
+ "learning_rate": 0.0002,
+ "loss": 0.3895,
+ "step": 1710
+ },
+ {
+ "epoch": 2.85,
+ "learning_rate": 0.0002,
+ "loss": 0.3689,
+ "step": 1720
+ },
+ {
+ "epoch": 2.87,
+ "learning_rate": 0.0002,
+ "loss": 0.3731,
+ "step": 1730
+ },
+ {
+ "epoch": 2.89,
+ "learning_rate": 0.0002,
+ "loss": 0.3839,
+ "step": 1740
+ },
+ {
+ "epoch": 2.9,
+ "learning_rate": 0.0002,
+ "loss": 0.3966,
+ "step": 1750
+ },
+ {
+ "epoch": 2.92,
+ "learning_rate": 0.0002,
+ "loss": 0.3828,
+ "step": 1760
+ },
+ {
+ "epoch": 2.94,
+ "learning_rate": 0.0002,
+ "loss": 0.3826,
+ "step": 1770
+ },
+ {
+ "epoch": 2.95,
+ "learning_rate": 0.0002,
+ "loss": 0.3842,
+ "step": 1780
+ },
+ {
+ "epoch": 2.97,
+ "learning_rate": 0.0002,
+ "loss": 0.3809,
+ "step": 1790
+ },
+ {
+ "epoch": 2.99,
+ "learning_rate": 0.0002,
+ "loss": 0.3923,
+ "step": 1800
+ },
+ {
+ "epoch": 2.99,
+ "eval_loss": 0.5423023700714111,
+ "eval_runtime": 241.4077,
+ "eval_samples_per_second": 4.142,
+ "eval_steps_per_second": 2.071,
+ "step": 1800
+ },
+ {
+ "epoch": 2.99,
+ "mmlu_eval_accuracy": 0.4921195253723927,
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
+ "mmlu_eval_accuracy_astronomy": 0.4375,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
+ "mmlu_eval_accuracy_college_biology": 0.5625,
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
+ "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
+ "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
+ "mmlu_eval_accuracy_econometrics": 0.08333333333333333,
+ "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.21951219512195122,
+ "mmlu_eval_accuracy_formal_logic": 0.5714285714285714,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.4375,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
+ "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
+ "mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
+ "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
+ "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+ "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+ "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+ "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
+ "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
+ "mmlu_eval_accuracy_moral_scenarios": 0.28,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
+ "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
+ "mmlu_eval_accuracy_professional_law": 0.34705882352941175,
+ "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
+ "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+ "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
+ "mmlu_eval_accuracy_sociology": 0.7272727272727273,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
+ "mmlu_eval_accuracy_virology": 0.3888888888888889,
+ "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
+ "mmlu_loss": 1.2815552267875747,
+ "step": 1800
+ }
+ ],
+ "max_steps": 5000,
+ "num_train_epochs": 9,
+ "total_flos": 4.58993116911403e+17,
+ "trial_name": null,
+ "trial_params": null
+}

checkpoint-1800/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d514a307ed83992d1218d4b4898c5c0916be1d496cbb3b3a2a7ec205916d1f96
+size 6011