Training in progress, step 400

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b99da48e7d4935b9282303bdaae1fc5be1d79368f183b104a2985a3dbda44098
+oid sha256:e364063e92c3c93a2dfffa029e7a03b5d1aabde442aeeafdc352f9ecd2d5378e
 size 319977229

checkpoint-200/adapter_model/adapter_model/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-200/adapter_model/adapter_model/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": [
+ "gate_proj",
+ "v_proj",
+ "down_proj",
+ "up_proj",
+ "o_proj",
+ "q_proj",
+ "k_proj"
+ ],
+ "task_type": "CAUSAL_LM"
+}

checkpoint-200/adapter_model/adapter_model/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b99da48e7d4935b9282303bdaae1fc5be1d79368f183b104a2985a3dbda44098
+size 319977229

checkpoint-400/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-400/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": [
+ "gate_proj",
+ "v_proj",
+ "down_proj",
+ "up_proj",
+ "o_proj",
+ "q_proj",
+ "k_proj"
+ ],
+ "task_type": "CAUSAL_LM"
+}

checkpoint-400/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e364063e92c3c93a2dfffa029e7a03b5d1aabde442aeeafdc352f9ecd2d5378e
+size 319977229

checkpoint-400/added_tokens.json

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

checkpoint-400/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16ee928388947e5e6577e0efe651c2187757b14c6453dba3b9d7bc3c3107ddde
+size 1279539973

checkpoint-400/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bceb192f2a338790ce6519cc6b7f9a104549523ed44eb87ca331d45f06cbcd78
+size 14511

checkpoint-400/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13276f15dd2b6acc19b970176aa2db4ac9b58241843e72c89b50e3094e903b19
+size 627

checkpoint-400/special_tokens_map.json

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

checkpoint-400/tokenizer.model

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

checkpoint-400/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-400/trainer_state.json

@@ -0,0 +1,398 @@
+{
+ "best_metric": 0.5613595843315125,
+ "best_model_checkpoint": "./output_v2/7b_cluster016_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_016/checkpoint-400",
+ "epoch": 0.6683375104427736,
+ "global_step": 400,
+ "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": 0.6709,
+ "step": 10
+ },
+ {
+ "epoch": 0.03,
+ "learning_rate": 0.0002,
+ "loss": 0.6086,
+ "step": 20
+ },
+ {
+ "epoch": 0.05,
+ "learning_rate": 0.0002,
+ "loss": 0.6457,
+ "step": 30
+ },
+ {
+ "epoch": 0.07,
+ "learning_rate": 0.0002,
+ "loss": 0.5722,
+ "step": 40
+ },
+ {
+ "epoch": 0.08,
+ "learning_rate": 0.0002,
+ "loss": 0.6039,
+ "step": 50
+ },
+ {
+ "epoch": 0.1,
+ "learning_rate": 0.0002,
+ "loss": 0.6079,
+ "step": 60
+ },
+ {
+ "epoch": 0.12,
+ "learning_rate": 0.0002,
+ "loss": 0.6261,
+ "step": 70
+ },
+ {
+ "epoch": 0.13,
+ "learning_rate": 0.0002,
+ "loss": 0.5876,
+ "step": 80
+ },
+ {
+ "epoch": 0.15,
+ "learning_rate": 0.0002,
+ "loss": 0.5645,
+ "step": 90
+ },
+ {
+ "epoch": 0.17,
+ "learning_rate": 0.0002,
+ "loss": 0.5902,
+ "step": 100
+ },
+ {
+ "epoch": 0.18,
+ "learning_rate": 0.0002,
+ "loss": 0.6008,
+ "step": 110
+ },
+ {
+ "epoch": 0.2,
+ "learning_rate": 0.0002,
+ "loss": 0.588,
+ "step": 120
+ },
+ {
+ "epoch": 0.22,
+ "learning_rate": 0.0002,
+ "loss": 0.6001,
+ "step": 130
+ },
+ {
+ "epoch": 0.23,
+ "learning_rate": 0.0002,
+ "loss": 0.5684,
+ "step": 140
+ },
+ {
+ "epoch": 0.25,
+ "learning_rate": 0.0002,
+ "loss": 0.5776,
+ "step": 150
+ },
+ {
+ "epoch": 0.27,
+ "learning_rate": 0.0002,
+ "loss": 0.5482,
+ "step": 160
+ },
+ {
+ "epoch": 0.28,
+ "learning_rate": 0.0002,
+ "loss": 0.5759,
+ "step": 170
+ },
+ {
+ "epoch": 0.3,
+ "learning_rate": 0.0002,
+ "loss": 0.5736,
+ "step": 180
+ },
+ {
+ "epoch": 0.32,
+ "learning_rate": 0.0002,
+ "loss": 0.5417,
+ "step": 190
+ },
+ {
+ "epoch": 0.33,
+ "learning_rate": 0.0002,
+ "loss": 0.5581,
+ "step": 200
+ },
+ {
+ "epoch": 0.33,
+ "eval_loss": 0.5835912227630615,
+ "eval_runtime": 210.8375,
+ "eval_samples_per_second": 4.743,
+ "eval_steps_per_second": 2.371,
+ "step": 200
+ },
+ {
+ "epoch": 0.33,
+ "mmlu_eval_accuracy": 0.46615455952712886,
+ "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+ "mmlu_eval_accuracy_astronomy": 0.4375,
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+ "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
+ "mmlu_eval_accuracy_college_biology": 0.5,
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
+ "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
+ "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+ "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
+ "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
+ "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
+ "mmlu_eval_accuracy_global_facts": 0.6,
+ "mmlu_eval_accuracy_high_school_biology": 0.34375,
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
+ "mmlu_eval_accuracy_high_school_european_history": 0.5,
+ "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
+ "mmlu_eval_accuracy_high_school_psychology": 0.7,
+ "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
+ "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5,
+ "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
+ "mmlu_eval_accuracy_international_law": 0.7692307692307693,
+ "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+ "mmlu_eval_accuracy_management": 0.45454545454545453,
+ "mmlu_eval_accuracy_marketing": 0.76,
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+ "mmlu_eval_accuracy_miscellaneous": 0.686046511627907,
+ "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
+ "mmlu_eval_accuracy_moral_scenarios": 0.23,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+ "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+ "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
+ "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
+ "mmlu_eval_accuracy_professional_medicine": 0.3870967741935484,
+ "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+ "mmlu_eval_accuracy_security_studies": 0.5925925925925926,
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.1704803334048148,
+ "step": 200
+ },
+ {
+ "epoch": 0.35,
+ "learning_rate": 0.0002,
+ "loss": 0.5413,
+ "step": 210
+ },
+ {
+ "epoch": 0.37,
+ "learning_rate": 0.0002,
+ "loss": 0.5632,
+ "step": 220
+ },
+ {
+ "epoch": 0.38,
+ "learning_rate": 0.0002,
+ "loss": 0.5752,
+ "step": 230
+ },
+ {
+ "epoch": 0.4,
+ "learning_rate": 0.0002,
+ "loss": 0.5583,
+ "step": 240
+ },
+ {
+ "epoch": 0.42,
+ "learning_rate": 0.0002,
+ "loss": 0.5274,
+ "step": 250
+ },
+ {
+ "epoch": 0.43,
+ "learning_rate": 0.0002,
+ "loss": 0.6163,
+ "step": 260
+ },
+ {
+ "epoch": 0.45,
+ "learning_rate": 0.0002,
+ "loss": 0.5673,
+ "step": 270
+ },
+ {
+ "epoch": 0.47,
+ "learning_rate": 0.0002,
+ "loss": 0.5241,
+ "step": 280
+ },
+ {
+ "epoch": 0.48,
+ "learning_rate": 0.0002,
+ "loss": 0.5564,
+ "step": 290
+ },
+ {
+ "epoch": 0.5,
+ "learning_rate": 0.0002,
+ "loss": 0.5451,
+ "step": 300
+ },
+ {
+ "epoch": 0.52,
+ "learning_rate": 0.0002,
+ "loss": 0.5661,
+ "step": 310
+ },
+ {
+ "epoch": 0.53,
+ "learning_rate": 0.0002,
+ "loss": 0.5467,
+ "step": 320
+ },
+ {
+ "epoch": 0.55,
+ "learning_rate": 0.0002,
+ "loss": 0.5513,
+ "step": 330
+ },
+ {
+ "epoch": 0.57,
+ "learning_rate": 0.0002,
+ "loss": 0.5999,
+ "step": 340
+ },
+ {
+ "epoch": 0.58,
+ "learning_rate": 0.0002,
+ "loss": 0.5518,
+ "step": 350
+ },
+ {
+ "epoch": 0.6,
+ "learning_rate": 0.0002,
+ "loss": 0.4896,
+ "step": 360
+ },
+ {
+ "epoch": 0.62,
+ "learning_rate": 0.0002,
+ "loss": 0.5314,
+ "step": 370
+ },
+ {
+ "epoch": 0.63,
+ "learning_rate": 0.0002,
+ "loss": 0.5495,
+ "step": 380
+ },
+ {
+ "epoch": 0.65,
+ "learning_rate": 0.0002,
+ "loss": 0.553,
+ "step": 390
+ },
+ {
+ "epoch": 0.67,
+ "learning_rate": 0.0002,
+ "loss": 0.5324,
+ "step": 400
+ },
+ {
+ "epoch": 0.67,
+ "eval_loss": 0.5613595843315125,
+ "eval_runtime": 210.7546,
+ "eval_samples_per_second": 4.745,
+ "eval_steps_per_second": 2.372,
+ "step": 400
+ },
+ {
+ "epoch": 0.67,
+ "mmlu_eval_accuracy": 0.47295524277138523,
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+ "mmlu_eval_accuracy_astronomy": 0.4375,
+ "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+ "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": 0.45454545454545453,
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
+ "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+ "mmlu_eval_accuracy_electrical_engineering": 0.4375,
+ "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
+ "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
+ "mmlu_eval_accuracy_global_facts": 0.5,
+ "mmlu_eval_accuracy_high_school_biology": 0.34375,
+ "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": 0.8181818181818182,
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
+ "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
+ "mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
+ "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
+ "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+ "mmlu_eval_accuracy_high_school_world_history": 0.5,
+ "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": 0.36363636363636365,
+ "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": 0.6744186046511628,
+ "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
+ "mmlu_eval_accuracy_moral_scenarios": 0.23,
+ "mmlu_eval_accuracy_nutrition": 0.6060606060606061,
+ "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
+ "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+ "mmlu_eval_accuracy_professional_accounting": 0.25806451612903225,
+ "mmlu_eval_accuracy_professional_law": 0.35294117647058826,
+ "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": 0.48148148148148145,
+ "mmlu_eval_accuracy_sociology": 0.6363636363636364,
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+ "mmlu_loss": 1.1208655159366037,
+ "step": 400
+ }
+ ],
+ "max_steps": 5000,
+ "num_train_epochs": 9,
+ "total_flos": 9.749315379481805e+16,
+ "trial_name": null,
+ "trial_params": null
+}

checkpoint-400/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b7337cac6f274c1c1c373fec2f0007d67f40ba802cf18675c9f0b0a46a5232b
+size 6011