Training in progress, step 400, checkpoint

checkpoint-400/README.md

@@ -0,0 +1,21 @@
+---
+library_name: peft
+---
+## Training procedure
+
+
+The following `bitsandbytes` quantization config was used during training:
+- quant_method: bitsandbytes
+- 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": "SkunkworksAI/Mistralic-7B-1",
+  "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",
+    "down_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-400/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:017c246d98fe9c51875e0eea827ca1fae2d8efaff082f922b7f5c96247abd88e
+size 335706314

checkpoint-400/adapter_model/adapter_model/README.md

@@ -0,0 +1,21 @@
+---
+library_name: peft
+---
+## Training procedure
+
+
+The following `bitsandbytes` quantization config was used during training:
+- quant_method: bitsandbytes
+- 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_model/adapter_model/adapter_config.json

@@ -0,0 +1,26 @@
+{
+  "auto_mapping": null,
+  "base_model_name_or_path": "SkunkworksAI/Mistralic-7B-1",
+  "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",
+    "down_proj",
+    "o_proj",
+    "gate_proj",
+    "up_proj",
+    "k_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-400/adapter_model/adapter_model/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:017c246d98fe9c51875e0eea827ca1fae2d8efaff082f922b7f5c96247abd88e
+size 335706314

checkpoint-400/added_tokens.json

@@ -0,0 +1,8 @@
+{
+  "</s>": 2,
+  "<s>": 1,
+  "<unk>": 0,
+  "<|im_end|>": 32000,
+  "<|im_start|>": 32001,
+  "[PAD]": 32002
+}

checkpoint-400/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e2576aa47b2182e8a6b7af56d97ce97cc202d386dc69ed278a2cf2d68b89620
+size 1342453434

checkpoint-400/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21f9d0a113f545acbb11408ed62e06d8fedcdc59f44df848ce3b18756be66cd8
+size 14180

checkpoint-400/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a3dedf52aaea6588501a76bbcc0db7c74d91e0049f19cb86f239096b664186d
+size 1064

checkpoint-400/special_tokens_map.json

@@ -0,0 +1,11 @@
+{
+  "additional_special_tokens": [
+    "<unk>",
+    "<s>",
+    "</s>"
+  ],
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "[PAD]",
+  "unk_token": "<unk>"
+}

checkpoint-400/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
+size 493443

checkpoint-400/tokenizer_config.json

@@ -0,0 +1,74 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    },
+    "32000": {
+      "content": "<|im_end|>",
+      "lstrip": true,
+      "normalized": true,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "32001": {
+      "content": "<|im_start|>",
+      "lstrip": true,
+      "normalized": true,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "32002": {
+      "content": "[PAD]",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<unk>",
+    "<s>",
+    "</s>"
+  ],
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "[PAD]",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "tokenizer_file": null,
+  "trust_remote_code": false,
+  "unk_token": "<unk>",
+  "use_default_system_prompt": true,
+  "use_fast": true
+}

checkpoint-400/trainer_state.json

@@ -0,0 +1,401 @@
+{
+  "best_metric": 0.7457320690155029,
+  "best_model_checkpoint": "experts/mistralic-expert-16/checkpoint-400",
+  "epoch": 0.1267427122940431,
+  "eval_steps": 200,
+  "global_step": 400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7873,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8128,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8641,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8246,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7867,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7705,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7671,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8725,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8337,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7819,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7729,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8169,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7988,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8958,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7682,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7729,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7375,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7756,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8256,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8504,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.7519773244857788,
+      "eval_runtime": 153.135,
+      "eval_samples_per_second": 6.53,
+      "eval_steps_per_second": 3.265,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "mmlu_eval_accuracy": 0.5986624199855438,
+      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
+      "mmlu_eval_accuracy_astronomy": 0.6875,
+      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
+      "mmlu_eval_accuracy_college_biology": 0.5625,
+      "mmlu_eval_accuracy_college_chemistry": 0.375,
+      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_mathematics": 0.6363636363636364,
+      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
+      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
+      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
+      "mmlu_eval_accuracy_electrical_engineering": 0.625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
+      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
+      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_high_school_biology": 0.59375,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
+      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
+      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
+      "mmlu_eval_accuracy_high_school_psychology": 0.85,
+      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5,
+      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
+      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 1.0,
+      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
+      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
+      "mmlu_eval_accuracy_moral_scenarios": 0.27,
+      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
+      "mmlu_eval_accuracy_philosophy": 0.7647058823529411,
+      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
+      "mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
+      "mmlu_eval_accuracy_professional_law": 0.4,
+      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
+      "mmlu_eval_accuracy_professional_psychology": 0.6231884057971014,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.6296296296296297,
+      "mmlu_eval_accuracy_sociology": 0.8181818181818182,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
+      "mmlu_eval_accuracy_virology": 0.5,
+      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
+      "mmlu_loss": 1.2932990876174781,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.823,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8241,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8277,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8068,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7698,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8068,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7913,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8086,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8127,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7804,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7667,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.757,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7438,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.768,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8151,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7718,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7903,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7447,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7712,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7808,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.7457320690155029,
+      "eval_runtime": 152.8304,
+      "eval_samples_per_second": 6.543,
+      "eval_steps_per_second": 3.272,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "mmlu_eval_accuracy": 0.5941853117334526,
+      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+      "mmlu_eval_accuracy_anatomy": 0.5,
+      "mmlu_eval_accuracy_astronomy": 0.75,
+      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
+      "mmlu_eval_accuracy_college_biology": 0.5625,
+      "mmlu_eval_accuracy_college_chemistry": 0.25,
+      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
+      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
+      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+      "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
+      "mmlu_eval_accuracy_econometrics": 0.5,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
+      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
+      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_high_school_biology": 0.625,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.6923076923076923,
+      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
+      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5,
+      "mmlu_eval_accuracy_international_law": 1.0,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
+      "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 1.0,
+      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
+      "mmlu_eval_accuracy_moral_disputes": 0.6052631578947368,
+      "mmlu_eval_accuracy_moral_scenarios": 0.32,
+      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
+      "mmlu_eval_accuracy_philosophy": 0.7058823529411765,
+      "mmlu_eval_accuracy_prehistory": 0.6285714285714286,
+      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
+      "mmlu_eval_accuracy_professional_law": 0.4294117647058823,
+      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
+      "mmlu_eval_accuracy_professional_psychology": 0.5797101449275363,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.5555555555555556,
+      "mmlu_eval_accuracy_sociology": 0.8181818181818182,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
+      "mmlu_eval_accuracy_virology": 0.3888888888888889,
+      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
+      "mmlu_loss": 1.0765447558606573,
+      "step": 400
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9468,
+  "num_train_epochs": 3,
+  "save_steps": 200,
+  "total_flos": 2.446105870364836e+17,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-400/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d0e0baa1bb2aa5937233bd631ead3549b3c1482ab16d538a9907b26a462c171
+size 6392