Model save

README.md

@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.0239
+- Loss: 0.9899
 
 ## Model description
 
@@ -40,8 +40,8 @@ The following hyperparameters were used during training:
 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 4
-- gradient_accumulation_steps: 256
-- total_train_batch_size: 2048
+- gradient_accumulation_steps: 64
+- total_train_batch_size: 512
 - total_eval_batch_size: 16
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
@@ -51,7 +51,7 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 1.0224        | 0.67  | 68   | 1.0240          |
+| 0.9866        | 0.67  | 272  | 0.9900          |
 
 
 ### Framework versions

adapter_config.json

@@ -16,10 +16,10 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
     "q_proj",
-    "o_proj",
-    "v_proj"
+    "v_proj",
+    "k_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9c0c9f426276e25b3a38839f3edf593998480ae8aaf9a7dd86f4091e745ac676
+oid sha256:3de510c187a9b9b133939f860696776514cf87b55aa2904f6ad68d714851f1d9
 size 218138576

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 0.67,
-    "eval_loss": 1.0239042043685913,
-    "eval_runtime": 2000.5663,
+    "eval_loss": 0.9899260401725769,
+    "eval_runtime": 2001.7837,
     "eval_samples": 23110,
-    "eval_samples_per_second": 11.552,
+    "eval_samples_per_second": 11.545,
     "eval_steps_per_second": 0.722,
-    "train_loss": 1.0572445532854866,
-    "train_runtime": 62451.9899,
+    "train_loss": 1.009627375532599,
+    "train_runtime": 62494.9184,
     "train_samples": 207865,
-    "train_samples_per_second": 3.328,
-    "train_steps_per_second": 0.002
+    "train_samples_per_second": 3.326,
+    "train_steps_per_second": 0.006
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 0.67,
-    "eval_loss": 1.0239042043685913,
-    "eval_runtime": 2000.5663,
+    "eval_loss": 0.9899260401725769,
+    "eval_runtime": 2001.7837,
     "eval_samples": 23110,
-    "eval_samples_per_second": 11.552,
+    "eval_samples_per_second": 11.545,
     "eval_steps_per_second": 0.722
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 0.67,
-    "train_loss": 1.0572445532854866,
-    "train_runtime": 62451.9899,
+    "train_loss": 1.009627375532599,
+    "train_runtime": 62494.9184,
     "train_samples": 207865,
-    "train_samples_per_second": 3.328,
-    "train_steps_per_second": 0.002
+    "train_samples_per_second": 3.326,
+    "train_steps_per_second": 0.006
 }

trainer_state.json

@@ -3,118 +3,364 @@
   "best_model_checkpoint": null,
   "epoch": 0.6699507389162561,
   "eval_steps": 500,
-  "global_step": 68,
+  "global_step": 272,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
-      "learning_rate": 1.999516282291988e-05,
-      "loss": 1.1268,
+      "epoch": 0.0,
+      "learning_rate": 1.9999700625010444e-05,
+      "loss": 1.1296,
       "step": 1
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.987930439740757e-05,
-      "loss": 1.1283,
+      "epoch": 0.01,
+      "learning_rate": 1.999251652147735e-05,
+      "loss": 1.12,
       "step": 5
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.95201310753273e-05,
-      "loss": 1.1092,
+      "epoch": 0.02,
+      "learning_rate": 1.997007728639956e-05,
+      "loss": 1.1064,
       "step": 10
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.8931150161867917e-05,
-      "loss": 1.0944,
+      "epoch": 0.04,
+      "learning_rate": 1.9932715879473385e-05,
+      "loss": 1.0915,
       "step": 15
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.8126579138282502e-05,
-      "loss": 1.0769,
+      "epoch": 0.05,
+      "learning_rate": 1.9880488219356086e-05,
+      "loss": 1.094,
       "step": 20
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.7125839641475074e-05,
+      "epoch": 0.06,
+      "learning_rate": 1.981347247496222e-05,
       "loss": 1.0658,
       "step": 25
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.595308864276666e-05,
-      "loss": 1.0595,
+      "epoch": 0.07,
+      "learning_rate": 1.973176894846855e-05,
+      "loss": 1.0527,
       "step": 30
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.4636635319853274e-05,
-      "loss": 1.0497,
+      "epoch": 0.09,
+      "learning_rate": 1.963549992519223e-05,
+      "loss": 1.0502,
       "step": 35
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.3208257698153677e-05,
-      "loss": 1.0332,
+      "epoch": 0.1,
+      "learning_rate": 1.9524809490566878e-05,
+      "loss": 1.0304,
       "step": 40
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.1702435557223988e-05,
-      "loss": 1.0312,
+      "epoch": 0.11,
+      "learning_rate": 1.939986331449053e-05,
+      "loss": 1.0267,
       "step": 45
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.0155518119203511e-05,
+      "epoch": 0.12,
+      "learning_rate": 1.926084840336821e-05,
       "loss": 1.029,
       "step": 50
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 8.604846610560771e-06,
-      "loss": 1.0258,
+      "epoch": 0.14,
+      "learning_rate": 1.910797282022027e-05,
+      "loss": 1.02,
       "step": 55
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 7.0878528777274814e-06,
-      "loss": 1.0258,
+      "epoch": 0.15,
+      "learning_rate": 1.894146537327533e-05,
+      "loss": 1.0127,
       "step": 60
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 5.64115581524629e-06,
-      "loss": 1.0224,
+      "epoch": 0.16,
+      "learning_rate": 1.8761575273514005e-05,
+      "loss": 1.0078,
       "step": 65
     },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8568571761675893e-05,
+      "loss": 1.0004,
+      "step": 70
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8362743705288127e-05,
+      "loss": 1.0081,
+      "step": 75
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814439916631857e-05,
+      "loss": 1.0157,
+      "step": 80
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791386494010081e-05,
+      "loss": 1.0118,
+      "step": 85
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7671486066220965e-05,
+      "loss": 1.0012,
+      "step": 90
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7417625312098453e-05,
+      "loss": 0.9919,
+      "step": 95
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7152662630033506e-05,
+      "loss": 1.0025,
+      "step": 100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6876994588534234e-05,
+      "loss": 1.0187,
+      "step": 105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659103377877423e-05,
+      "loss": 0.9959,
+      "step": 110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629520819706912e-05,
+      "loss": 0.9937,
+      "step": 115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.598996060429634e-05,
+      "loss": 0.9981,
+      "step": 120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.56757478632168e-05,
+      "loss": 0.9954,
+      "step": 125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5353040254690396e-05,
+      "loss": 1.0044,
+      "step": 130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5022320773808612e-05,
+      "loss": 1.0033,
+      "step": 135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 0.9901,
+      "step": 140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.9965,
+      "step": 145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3987096456067236e-05,
+      "loss": 0.9915,
+      "step": 150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3629388050819547e-05,
+      "loss": 0.9862,
+      "step": 155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3266247556066122e-05,
+      "loss": 0.9757,
+      "step": 160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2898218482625606e-05,
+      "loss": 0.9944,
+      "step": 165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.252585165803135e-05,
+      "loss": 0.9963,
+      "step": 170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2149704402110243e-05,
+      "loss": 0.9818,
+      "step": 175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1770339692844484e-05,
+      "loss": 0.9881,
+      "step": 180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1388325323764889e-05,
+      "loss": 0.9965,
+      "step": 185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1004233054136726e-05,
+      "loss": 0.9922,
+      "step": 190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0618637753210086e-05,
+      "loss": 0.9934,
+      "step": 195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0232116539815558e-05,
+      "loss": 0.9823,
+      "step": 200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.845247918592937e-06,
+      "loss": 0.9881,
+      "step": 205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.458610914145826e-06,
+      "loss": 0.9867,
+      "step": 210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.072784204417995e-06,
+      "loss": 0.9983,
+      "step": 215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.688345254588579e-06,
+      "loss": 0.9866,
+      "step": 220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.305869452782446e-06,
+      "loss": 0.9918,
+      "step": 225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.92592924888925e-06,
+      "loss": 0.9902,
+      "step": 230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.549093297780133e-06,
+      "loss": 0.987,
+      "step": 235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.175925608204428e-06,
+      "loss": 0.9941,
+      "step": 240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.806984698640202e-06,
+      "loss": 0.9968,
+      "step": 245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.442822761362015e-06,
+      "loss": 0.9842,
+      "step": 250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.083984835977154e-06,
+      "loss": 0.9788,
+      "step": 255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.731007993667155e-06,
+      "loss": 0.993,
+      "step": 260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.38442053335571e-06,
+      "loss": 0.9834,
+      "step": 265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.044741191005908e-06,
+      "loss": 0.9866,
+      "step": 270
+    },
     {
       "epoch": 0.67,
-      "eval_loss": 1.0239689350128174,
-      "eval_runtime": 2002.0819,
-      "eval_samples_per_second": 11.543,
+      "eval_loss": 0.9899875521659851,
+      "eval_runtime": 2002.7249,
+      "eval_samples_per_second": 11.539,
       "eval_steps_per_second": 0.722,
-      "step": 68
+      "step": 272
     },
     {
       "epoch": 0.67,
-      "step": 68,
-      "total_flos": 1.2276051069330522e+19,
-      "train_loss": 1.0572445532854866,
-      "train_runtime": 62451.9899,
-      "train_samples_per_second": 3.328,
-      "train_steps_per_second": 0.002
+      "step": 272,
+      "total_flos": 1.2276051112280195e+19,
+      "train_loss": 1.009627375532599,
+      "train_runtime": 62494.9184,
+      "train_samples_per_second": 3.326,
+      "train_steps_per_second": 0.006
     }
   ],
   "logging_steps": 5,
-  "max_steps": 101,
+  "max_steps": 406,
   "num_train_epochs": 1,
   "save_steps": 500,
-  "total_flos": 1.2276051069330522e+19,
+  "total_flos": 1.2276051112280195e+19,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:36facd59ad2755544b27eb1044191870ef084d9fb2b53ff4c6f47efcc03ef423
+oid sha256:1f7940da9943c63332d8a273f9889b5a0eb3d404a4cf0c9b9884f3e587c5229d
 size 4155