Training in progress, step 400

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dbaccf9d57b49b148f25beb4fefd5a014a14a88d2d0f184bc7971e53c80a3bfa
+oid sha256:e46fccfc2b3bca2deec2f0ce8c8a622f863ee1444c4cdb9002c588fc8cb5574f
 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": [
+    "v_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj",
+    "up_proj",
+    "gate_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:dbaccf9d57b49b148f25beb4fefd5a014a14a88d2d0f184bc7971e53c80a3bfa
+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": [
+    "v_proj",
+    "o_proj",
+    "k_proj",
+    "down_proj",
+    "q_proj",
+    "up_proj",
+    "gate_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

checkpoint-400/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e46fccfc2b3bca2deec2f0ce8c8a622f863ee1444c4cdb9002c588fc8cb5574f
+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:c37450b0a4d03aedb9572c247cbb3231a93cc1f35dc19fd8a9d9aa0c556c588a
+size 1279539973

checkpoint-400/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9003963d891adc27fcd80dea3775b2f03a25bad1740507deb401af323ae41825
+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": 0.5141976475715637,
+  "best_model_checkpoint": "./output_v2/7b_cluster05_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_05/checkpoint-400",
+  "epoch": 1.1895910780669146,
+  "global_step": 400,
+  "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": 0.6832,
+      "step": 10
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5389,
+      "step": 20
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5072,
+      "step": 30
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5211,
+      "step": 40
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.5888,
+      "step": 50
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.534,
+      "step": 60
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.5684,
+      "step": 70
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002,
+      "loss": 0.514,
+      "step": 80
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002,
+      "loss": 0.5518,
+      "step": 90
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002,
+      "loss": 0.5034,
+      "step": 100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002,
+      "loss": 0.542,
+      "step": 110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002,
+      "loss": 0.4847,
+      "step": 120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002,
+      "loss": 0.4772,
+      "step": 130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002,
+      "loss": 0.5196,
+      "step": 140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002,
+      "loss": 0.4672,
+      "step": 150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002,
+      "loss": 0.4913,
+      "step": 160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002,
+      "loss": 0.5498,
+      "step": 170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002,
+      "loss": 0.5328,
+      "step": 180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002,
+      "loss": 0.5313,
+      "step": 190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002,
+      "loss": 0.515,
+      "step": 200
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.531296968460083,
+      "eval_runtime": 174.4771,
+      "eval_samples_per_second": 5.731,
+      "eval_steps_per_second": 2.866,
+      "step": 200
+    },
+    {
+      "epoch": 0.59,
+      "mmlu_eval_accuracy": 0.46292469330066577,
+      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
+      "mmlu_eval_accuracy_anatomy": 0.5,
+      "mmlu_eval_accuracy_astronomy": 0.5,
+      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
+      "mmlu_eval_accuracy_college_biology": 0.3125,
+      "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": 0.3181818181818182,
+      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+      "mmlu_eval_accuracy_computer_security": 0.18181818181818182,
+      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
+      "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": 0.45454545454545453,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
+      "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": 0.20689655172413793,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.38461538461538464,
+      "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
+      "mmlu_eval_accuracy_high_school_psychology": 0.65,
+      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_world_history": 0.46153846153846156,
+      "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
+      "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
+      "mmlu_eval_accuracy_logical_fallacies": 0.5,
+      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+      "mmlu_eval_accuracy_management": 0.6363636363636364,
+      "mmlu_eval_accuracy_marketing": 0.72,
+      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+      "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
+      "mmlu_eval_accuracy_moral_disputes": 0.39473684210526316,
+      "mmlu_eval_accuracy_moral_scenarios": 0.24,
+      "mmlu_eval_accuracy_nutrition": 0.5454545454545454,
+      "mmlu_eval_accuracy_philosophy": 0.5,
+      "mmlu_eval_accuracy_prehistory": 0.4,
+      "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
+      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
+      "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
+      "mmlu_eval_accuracy_professional_psychology": 0.42028985507246375,
+      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
+      "mmlu_eval_accuracy_sociology": 0.5909090909090909,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
+      "mmlu_eval_accuracy_virology": 0.3888888888888889,
+      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+      "mmlu_loss": 1.3857738420406147,
+      "step": 200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002,
+      "loss": 0.4582,
+      "step": 210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002,
+      "loss": 0.4796,
+      "step": 220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002,
+      "loss": 0.5081,
+      "step": 230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002,
+      "loss": 0.5206,
+      "step": 240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002,
+      "loss": 0.4961,
+      "step": 250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002,
+      "loss": 0.5219,
+      "step": 260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002,
+      "loss": 0.5311,
+      "step": 270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002,
+      "loss": 0.5039,
+      "step": 280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002,
+      "loss": 0.4994,
+      "step": 290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002,
+      "loss": 0.4804,
+      "step": 300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002,
+      "loss": 0.4791,
+      "step": 310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002,
+      "loss": 0.4822,
+      "step": 320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002,
+      "loss": 0.5157,
+      "step": 330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5184,
+      "step": 340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4562,
+      "step": 350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4117,
+      "step": 360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002,
+      "loss": 0.451,
+      "step": 370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4237,
+      "step": 380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002,
+      "loss": 0.4243,
+      "step": 390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002,
+      "loss": 0.4409,
+      "step": 400
+    },
+    {
+      "epoch": 1.19,
+      "eval_loss": 0.5141976475715637,
+      "eval_runtime": 174.8657,
+      "eval_samples_per_second": 5.719,
+      "eval_steps_per_second": 2.859,
+      "step": 400
+    },
+    {
+      "epoch": 1.19,
+      "mmlu_eval_accuracy": 0.45097298030174665,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.5,
+      "mmlu_eval_accuracy_astronomy": 0.5,
+      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.41379310344827586,
+      "mmlu_eval_accuracy_college_biology": 0.4375,
+      "mmlu_eval_accuracy_college_chemistry": 0.0,
+      "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": 0.5454545454545454,
+      "mmlu_eval_accuracy_computer_security": 0.2727272727272727,
+      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+      "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
+      "mmlu_eval_accuracy_electrical_engineering": 0.3125,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
+      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
+      "mmlu_eval_accuracy_global_facts": 0.6,
+      "mmlu_eval_accuracy_high_school_biology": 0.3125,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
+      "mmlu_eval_accuracy_high_school_european_history": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.3488372093023256,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
+      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
+      "mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333,
+      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
+      "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": 0.7692307692307693,
+      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
+      "mmlu_eval_accuracy_logical_fallacies": 0.5,
+      "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
+      "mmlu_eval_accuracy_management": 0.5454545454545454,
+      "mmlu_eval_accuracy_marketing": 0.68,
+      "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": 0.24,
+      "mmlu_eval_accuracy_nutrition": 0.5151515151515151,
+      "mmlu_eval_accuracy_philosophy": 0.47058823529411764,
+      "mmlu_eval_accuracy_prehistory": 0.45714285714285713,
+      "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": 0.391304347826087,
+      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
+      "mmlu_eval_accuracy_sociology": 0.6363636363636364,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
+      "mmlu_eval_accuracy_virology": 0.3888888888888889,
+      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+      "mmlu_loss": 1.0380505950970687,
+      "step": 400
+    }
+  ],
+  "max_steps": 5000,
+  "num_train_epochs": 15,
+  "total_flos": 8.168382747058176e+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 sha256:1bcfffed19c275d0ca80136a02b48634f34069fc2c974a4ef89e0cd1066213ed
+size 6011