Training in progress, step 600

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cec13cca0e3fa9f9a8cf309e53b807f9b1d1fa31743e030e0ab167b39da06f9a
+oid sha256:a8f8fb017e765a52f28242fb4e74494f2e6389682a6d4764265e853e317716cb
 size 319977229

checkpoint-200/adapter_model/adapter_model/README.md

@@ -4,6 +4,17 @@ 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
+
 The following `bitsandbytes` quantization config was used during training:
 - load_in_8bit: False
 - load_in_4bit: True
@@ -16,5 +27,6 @@ The following `bitsandbytes` quantization config was used during training:
 - bnb_4bit_compute_dtype: bfloat16
 ### Framework versions
 
+- PEFT 0.4.0
 
 - PEFT 0.4.0

checkpoint-200/adapter_model/adapter_model/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4eed09b6f9c85e405d2b0e756af6958474fffba62098ec8b2bb259e6e13b0b79
+oid sha256:cec13cca0e3fa9f9a8cf309e53b807f9b1d1fa31743e030e0ab167b39da06f9a
 size 319977229

checkpoint-600/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-600/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": [
+    "o_proj",
+    "k_proj",
+    "q_proj",
+    "gate_proj",
+    "v_proj",
+    "up_proj",
+    "down_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-600/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8f8fb017e765a52f28242fb4e74494f2e6389682a6d4764265e853e317716cb
+size 319977229

checkpoint-600/added_tokens.json

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

checkpoint-600/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58e56065349bc903b89817f577de78341b20edad444d380c4050dbac63be9c99
+size 1279539973

checkpoint-600/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:432472437ce2285a9330dc4f17cbadb580403c4b9b3cb996fdc5727e215dfb60
+size 14511

checkpoint-600/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c1a2ac1c11599601eeac95feb1dbfd49ec5c625e61dcce18b3f094491f9cf2d
+size 627

checkpoint-600/special_tokens_map.json

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

checkpoint-600/tokenizer.model

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

checkpoint-600/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-600/trainer_state.json

@@ -0,0 +1,589 @@
+{
+  "best_metric": 0.32424649596214294,
+  "best_model_checkpoint": "./output_v2/7b_cluster03_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_03/checkpoint-200",
+  "epoch": 2.060085836909871,
+  "global_step": 600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8627,
+      "step": 10
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4107,
+      "step": 20
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3691,
+      "step": 30
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3979,
+      "step": 40
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.3495,
+      "step": 50
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.3693,
+      "step": 60
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002,
+      "loss": 0.357,
+      "step": 70
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002,
+      "loss": 0.3456,
+      "step": 80
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002,
+      "loss": 0.3122,
+      "step": 90
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002,
+      "loss": 0.3131,
+      "step": 100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002,
+      "loss": 0.3678,
+      "step": 110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002,
+      "loss": 0.3364,
+      "step": 120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002,
+      "loss": 0.324,
+      "step": 130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002,
+      "loss": 0.353,
+      "step": 140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002,
+      "loss": 0.3473,
+      "step": 150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002,
+      "loss": 0.363,
+      "step": 160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002,
+      "loss": 0.3192,
+      "step": 170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002,
+      "loss": 0.3125,
+      "step": 180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002,
+      "loss": 0.3001,
+      "step": 190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002,
+      "loss": 0.3164,
+      "step": 200
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.32424649596214294,
+      "eval_runtime": 166.2846,
+      "eval_samples_per_second": 6.014,
+      "eval_steps_per_second": 3.007,
+      "step": 200
+    },
+    {
+      "epoch": 0.69,
+      "mmlu_eval_accuracy": 0.46206132856742516,
+      "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
+      "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
+      "mmlu_eval_accuracy_astronomy": 0.4375,
+      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
+      "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": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
+      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
+      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
+      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+      "mmlu_eval_accuracy_econometrics": 0.25,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.2926829268292683,
+      "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": 0.3181818181818182,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_european_history": 0.5,
+      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.4230769230769231,
+      "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": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
+      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
+      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
+      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
+      "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
+      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+      "mmlu_eval_accuracy_management": 0.36363636363636365,
+      "mmlu_eval_accuracy_marketing": 0.72,
+      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
+      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
+      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
+      "mmlu_eval_accuracy_moral_scenarios": 0.24,
+      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
+      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
+      "mmlu_eval_accuracy_professional_law": 0.3,
+      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
+      "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
+      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
+      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
+      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
+      "mmlu_eval_accuracy_virology": 0.3888888888888889,
+      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+      "mmlu_loss": 0.954094658943754,
+      "step": 200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002,
+      "loss": 0.3516,
+      "step": 210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002,
+      "loss": 0.2846,
+      "step": 220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002,
+      "loss": 0.3048,
+      "step": 230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002,
+      "loss": 0.315,
+      "step": 240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002,
+      "loss": 0.3229,
+      "step": 250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002,
+      "loss": 0.3218,
+      "step": 260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002,
+      "loss": 0.3188,
+      "step": 270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002,
+      "loss": 0.3246,
+      "step": 280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002,
+      "loss": 0.2986,
+      "step": 290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002,
+      "loss": 0.2905,
+      "step": 300
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002,
+      "loss": 0.2847,
+      "step": 310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002,
+      "loss": 0.2384,
+      "step": 320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002,
+      "loss": 0.2815,
+      "step": 330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002,
+      "loss": 0.2737,
+      "step": 340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002,
+      "loss": 0.2985,
+      "step": 350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002,
+      "loss": 0.2747,
+      "step": 360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002,
+      "loss": 0.2691,
+      "step": 370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002,
+      "loss": 0.2823,
+      "step": 380
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002,
+      "loss": 0.2389,
+      "step": 390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002,
+      "loss": 0.2552,
+      "step": 400
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 0.3267834484577179,
+      "eval_runtime": 166.4022,
+      "eval_samples_per_second": 6.01,
+      "eval_steps_per_second": 3.005,
+      "step": 400
+    },
+    {
+      "epoch": 1.37,
+      "mmlu_eval_accuracy": 0.4687168428980155,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+      "mmlu_eval_accuracy_astronomy": 0.4375,
+      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
+      "mmlu_eval_accuracy_college_biology": 0.375,
+      "mmlu_eval_accuracy_college_chemistry": 0.375,
+      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
+      "mmlu_eval_accuracy_college_medicine": 0.2727272727272727,
+      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
+      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
+      "mmlu_eval_accuracy_econometrics": 0.25,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+      "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": 0.40625,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
+      "mmlu_eval_accuracy_high_school_european_history": 0.5,
+      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.37209302325581395,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
+      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8,
+      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
+      "mmlu_eval_accuracy_human_aging": 0.6086956521739131,
+      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+      "mmlu_eval_accuracy_management": 0.45454545454545453,
+      "mmlu_eval_accuracy_marketing": 0.72,
+      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
+      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
+      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
+      "mmlu_eval_accuracy_moral_scenarios": 0.24,
+      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
+      "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.1935483870967742,
+      "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
+      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
+      "mmlu_eval_accuracy_professional_psychology": 0.36231884057971014,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
+      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
+      "mmlu_eval_accuracy_virology": 0.4444444444444444,
+      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+      "mmlu_loss": 0.9237010620787311,
+      "step": 400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002,
+      "loss": 0.2629,
+      "step": 410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002,
+      "loss": 0.2455,
+      "step": 420
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002,
+      "loss": 0.2857,
+      "step": 430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002,
+      "loss": 0.2728,
+      "step": 440
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002,
+      "loss": 0.2814,
+      "step": 450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002,
+      "loss": 0.2993,
+      "step": 460
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002,
+      "loss": 0.282,
+      "step": 470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002,
+      "loss": 0.269,
+      "step": 480
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002,
+      "loss": 0.2534,
+      "step": 490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002,
+      "loss": 0.2965,
+      "step": 500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002,
+      "loss": 0.2872,
+      "step": 510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002,
+      "loss": 0.2742,
+      "step": 520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002,
+      "loss": 0.2568,
+      "step": 530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002,
+      "loss": 0.2722,
+      "step": 540
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002,
+      "loss": 0.2502,
+      "step": 550
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002,
+      "loss": 0.2755,
+      "step": 560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002,
+      "loss": 0.2695,
+      "step": 570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002,
+      "loss": 0.281,
+      "step": 580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002,
+      "loss": 0.2127,
+      "step": 590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002,
+      "loss": 0.1943,
+      "step": 600
+    },
+    {
+      "epoch": 2.06,
+      "eval_loss": 0.3318677246570587,
+      "eval_runtime": 166.3325,
+      "eval_samples_per_second": 6.012,
+      "eval_steps_per_second": 3.006,
+      "step": 600
+    },
+    {
+      "epoch": 2.06,
+      "mmlu_eval_accuracy": 0.463905720895551,
+      "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
+      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+      "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": 0.25,
+      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
+      "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
+      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
+      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+      "mmlu_eval_accuracy_econometrics": 0.25,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.2682926829268293,
+      "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": 0.3181818181818182,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
+      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.5714285714285714,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+      "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": 0.43478260869565216,
+      "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": 0.4166666666666667,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.2727272727272727,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+      "mmlu_eval_accuracy_management": 0.7272727272727273,
+      "mmlu_eval_accuracy_marketing": 0.72,
+      "mmlu_eval_accuracy_medical_genetics": 0.8181818181818182,
+      "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
+      "mmlu_eval_accuracy_moral_disputes": 0.42105263157894735,
+      "mmlu_eval_accuracy_moral_scenarios": 0.23,
+      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
+      "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
+      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
+      "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
+      "mmlu_eval_accuracy_professional_psychology": 0.37681159420289856,
+      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
+      "mmlu_eval_accuracy_security_studies": 0.4444444444444444,
+      "mmlu_eval_accuracy_sociology": 0.7272727272727273,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
+      "mmlu_eval_accuracy_virology": 0.4444444444444444,
+      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+      "mmlu_loss": 0.8365475967156825,
+      "step": 600
+    }
+  ],
+  "max_steps": 5000,
+  "num_train_epochs": 18,
+  "total_flos": 2.8797865278160896e+16,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-600/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bff4af165f56bb71101d04f975a703b5813501ad63fa33abd20808564b39893
+size 6011