Model save

README.md

@@ -0,0 +1,58 @@
+---
+base_model: hllj/mistral-vi-math
+tags:
+- generated_from_trainer
+model-index:
+- name: sft-mistral-v2-rank-64-alpha-128
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# sft-mistral-v2-rank-64-alpha-128
+
+This model is a fine-tuned version of [hllj/mistral-vi-math](https://huggingface.co/hllj/mistral-vi-math) on an unknown dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.4965
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 5e-05
+- train_batch_size: 8
+- eval_batch_size: 8
+- seed: 42
+- distributed_type: multi-GPU
+- gradient_accumulation_steps: 2
+- total_train_batch_size: 16
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.05
+- num_epochs: 2
+- mixed_precision_training: Native AMP
+
+### Training results
+
+
+
+### Framework versions
+
+- Transformers 4.35.2
+- Pytorch 2.1.0
+- Datasets 2.15.0
+- Tokenizers 0.15.0

adapter_config.json

@@ -0,0 +1,25 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": null,
+  "base_model_name_or_path": "hllj/mistral-vi-math",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "lora_alpha": 128,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 64,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "q_proj",
+    "o_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:585f67a6b98fe9814d09e92c9151e25abfcc00fc86b0481098669a6db9d2098b
+size 218138576

all_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 1.45,
+    "eval_loss": 0.4965249300003052,
+    "eval_runtime": 80.8754,
+    "eval_samples": 852,
+    "eval_samples_per_second": 10.535,
+    "eval_steps_per_second": 1.323,
+    "train_loss": 0.33035796076752416,
+    "train_runtime": 4854.2663,
+    "train_samples": 7665,
+    "train_samples_per_second": 3.158,
+    "train_steps_per_second": 0.197
+}

config_argument.yaml

@@ -0,0 +1,50 @@
+cache_dir: ./cache
+ddp_find_unused_parameters: false
+ddp_timeout: 30000
+device_map: auto
+do_eval: true
+do_train: true
+eval_steps: 500
+evaluation_strategy: steps
+fp16: true
+gradient_accumulation_steps: 2
+gradient_checkpointing: true
+gradient_checkpointing_kwargs:
+  use_reentrant: false
+hub_model_id: hllj/sft-mistral-v2-rank-64-alpha-128
+hub_strategy: every_save
+learning_rate: 5.0e-05
+log_level: info
+logging_first_step: true
+logging_steps: 10
+logging_strategy: steps
+lora_alpha: 128
+lora_dropout: 0.05
+lora_r: 64
+lora_target_modules:
+- q_proj
+- k_proj
+- v_proj
+- o_proj
+lr_scheduler_type: cosine
+max_seq_length: 1024
+model_name_or_path: hllj/mistral-vi-math
+model_type: auto
+num_train_epochs: 2
+output_dir: outputs-sft-mistral-v2-rank-64-alpha-128
+overwrite_output_dir: true
+per_device_eval_batch_size: 8
+per_device_train_batch_size: 8
+preprocessing_num_workers: 4
+push_to_hub: true
+report_to: wandb
+run_name: sft-mistral-v2-rank-64-alpha-128
+save_steps: 500
+save_strategy: steps
+save_total_limit: 13
+seed: 42
+torch_dtype: float16
+train_file_dir: datasets/finetune
+use_peft: true
+warmup_ratio: 0.05
+weight_decay: 0.05

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 1.45,
+    "eval_loss": 0.4965249300003052,
+    "eval_runtime": 80.8754,
+    "eval_samples": 852,
+    "eval_samples_per_second": 10.535,
+    "eval_steps_per_second": 1.323
+}

special_tokens_map.json

@@ -0,0 +1,24 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": "<unk>",
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

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

tokenizer_config.json

@@ -0,0 +1,43 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "legacy": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<unk>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": true
+}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 1.45,
+    "train_loss": 0.33035796076752416,
+    "train_runtime": 4854.2663,
+    "train_samples": 7665,
+    "train_samples_per_second": 3.158,
+    "train_steps_per_second": 0.197
+}

trainer_state.json

@@ -0,0 +1,292 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.448383733055266,
+  "eval_steps": 500,
+  "global_step": 430,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0416666666666667e-06,
+      "loss": 0.7671,
+      "step": 1
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.6676,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.5586,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.125e-05,
+      "loss": 0.4224,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.3747,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999940408373438e-05,
+      "loss": 0.3632,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.997854999726187e-05,
+      "loss": 0.3615,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.99279285007165e-05,
+      "loss": 0.346,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.984759992069235e-05,
+      "loss": 0.3425,
+      "step": 80
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.973765998627628e-05,
+      "loss": 0.3341,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.959823971496574e-05,
+      "loss": 0.3289,
+      "step": 100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.942950525653282e-05,
+      "loss": 0.3351,
+      "step": 110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.923165769502033e-05,
+      "loss": 0.3279,
+      "step": 120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.900493280910623e-05,
+      "loss": 0.3255,
+      "step": 130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.874960079112168e-05,
+      "loss": 0.3365,
+      "step": 140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.846596592505776e-05,
+      "loss": 0.3486,
+      "step": 150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.815436622394441e-05,
+      "loss": 0.3294,
+      "step": 160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.781517302703405e-05,
+      "loss": 0.3316,
+      "step": 170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.744879055726944e-05,
+      "loss": 0.3176,
+      "step": 180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.7055655439563656e-05,
+      "loss": 0.327,
+      "step": 190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6636236180465876e-05,
+      "loss": 0.3203,
+      "step": 200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.619103260983328e-05,
+      "loss": 0.3225,
+      "step": 210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.572057528517441e-05,
+      "loss": 0.3125,
+      "step": 220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.522542485937369e-05,
+      "loss": 0.3141,
+      "step": 230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4706171412550815e-05,
+      "loss": 0.2982,
+      "step": 240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.416343374885116e-05,
+      "loss": 0.2867,
+      "step": 250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.359785865900513e-05,
+      "loss": 0.2975,
+      "step": 260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.301012014953544e-05,
+      "loss": 0.3021,
+      "step": 270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.240091863953084e-05,
+      "loss": 0.3082,
+      "step": 280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.177098012594341e-05,
+      "loss": 0.306,
+      "step": 290
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.1121055318404264e-05,
+      "loss": 0.2859,
+      "step": 300
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.045191874458876e-05,
+      "loss": 0.2982,
+      "step": 310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.976436782719713e-05,
+      "loss": 0.2956,
+      "step": 320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.905922193365089e-05,
+      "loss": 0.2958,
+      "step": 330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8337321399637125e-05,
+      "loss": 0.2952,
+      "step": 340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.75995265276646e-05,
+      "loss": 0.2918,
+      "step": 350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.6846716561824965e-05,
+      "loss": 0.2909,
+      "step": 360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6079788639981036e-05,
+      "loss": 0.2927,
+      "step": 370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.529965672463056e-05,
+      "loss": 0.2889,
+      "step": 380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.450725051371987e-05,
+      "loss": 0.2752,
+      "step": 390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.370351433270529e-05,
+      "loss": 0.2883,
+      "step": 400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.288940600918266e-05,
+      "loss": 0.2761,
+      "step": 410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.2065895731425995e-05,
+      "loss": 0.2824,
+      "step": 420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.1233964892195884e-05,
+      "loss": 0.2919,
+      "step": 430
+    },
+    {
+      "epoch": 1.45,
+      "step": 430,
+      "total_flos": 3.022612632268964e+17,
+      "train_loss": 0.33035796076752416,
+      "train_runtime": 4854.2663,
+      "train_samples_per_second": 3.158,
+      "train_steps_per_second": 0.197
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 958,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 3.022612632268964e+17,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d1d023ff15474507d51f7190d602417774ec3256dd9c52464567a168763c921
+size 4664