Model save

README.md

@@ -2,13 +2,9 @@
 license: other
 library_name: peft
 tags:
-- alignment-handbook
-- generated_from_trainer
 - trl
 - sft
 - generated_from_trainer
-datasets:
-- ruozhiba
 base_model: 01-ai/Yi-6B
 model-index:
 - name: Yi-6B-ruozhiba
@@ -20,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # Yi-6B-ruozhiba
 
-This model is a fine-tuned version of [01-ai/Yi-6B](https://huggingface.co/01-ai/Yi-6B) on the ruozhiba dataset.
+This model is a fine-tuned version of [01-ai/Yi-6B](https://huggingface.co/01-ai/Yi-6B) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 4.1664
+- Loss: 1.8389
 
 ## Model description
 
@@ -48,32 +44,16 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 20
+- num_epochs: 4
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 2.333         | 1.0   | 55   | 2.0215          |
-| 1.8575        | 2.0   | 110  | 1.8371          |
-| 1.6435        | 3.0   | 165  | 1.8182          |
-| 1.4138        | 4.0   | 220  | 1.9196          |
-| 1.0749        | 5.0   | 275  | 2.2265          |
-| 0.6715        | 6.0   | 330  | 2.5187          |
-| 0.4252        | 7.0   | 385  | 2.8304          |
-| 0.2572        | 8.0   | 440  | 3.1702          |
-| 0.1543        | 9.0   | 495  | 3.3739          |
-| 0.1095        | 10.0  | 550  | 3.4964          |
-| 0.0875        | 11.0  | 605  | 3.6468          |
-| 0.0692        | 12.0  | 660  | 3.8002          |
-| 0.0573        | 13.0  | 715  | 3.9182          |
-| 0.0527        | 14.0  | 770  | 4.0117          |
-| 0.0494        | 15.0  | 825  | 4.0920          |
-| 0.0459        | 16.0  | 880  | 4.1272          |
-| 0.0456        | 17.0  | 935  | 4.1514          |
-| 0.0435        | 18.0  | 990  | 4.1618          |
-| 0.0446        | 19.0  | 1045 | 4.1660          |
-| 0.0428        | 20.0  | 1100 | 4.1664          |
+| 1.9256        | 1.0   | 55   | 1.8785          |
+| 1.6738        | 2.0   | 110  | 1.8229          |
+| 1.6181        | 3.0   | 165  | 1.8283          |
+| 1.5007        | 4.0   | 220  | 1.8389          |
 
 
 ### Framework versions

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 20.0,
-    "eval_loss": 4.166376113891602,
-    "eval_runtime": 4.9235,
+    "epoch": 4.0,
+    "eval_loss": 1.8388779163360596,
+    "eval_runtime": 4.9283,
     "eval_samples": 23,
-    "eval_samples_per_second": 4.672,
-    "eval_steps_per_second": 1.219,
-    "train_loss": 0.5259156136621128,
-    "train_runtime": 7800.8143,
+    "eval_samples_per_second": 4.667,
+    "eval_steps_per_second": 1.217,
+    "train_loss": 1.7501471573656255,
+    "train_runtime": 1644.9971,
     "train_samples": 217,
-    "train_samples_per_second": 0.556,
-    "train_steps_per_second": 0.141
+    "train_samples_per_second": 0.528,
+    "train_steps_per_second": 0.134
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 20.0,
-    "eval_loss": 4.166376113891602,
-    "eval_runtime": 4.9235,
+    "epoch": 4.0,
+    "eval_loss": 1.8388779163360596,
+    "eval_runtime": 4.9283,
     "eval_samples": 23,
-    "eval_samples_per_second": 4.672,
-    "eval_steps_per_second": 1.219
+    "eval_samples_per_second": 4.667,
+    "eval_steps_per_second": 1.217
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 20.0,
-    "train_loss": 0.5259156136621128,
-    "train_runtime": 7800.8143,
+    "epoch": 4.0,
+    "train_loss": 1.7501471573656255,
+    "train_runtime": 1644.9971,
     "train_samples": 217,
-    "train_samples_per_second": 0.556,
-    "train_steps_per_second": 0.141
+    "train_samples_per_second": 0.528,
+    "train_steps_per_second": 0.134
 }

trainer_state.json

@@ -1,625 +1,417 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 20.0,
+  "epoch": 4.0,
   "eval_steps": 500,
-  "global_step": 1100,
+  "global_step": 220,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.02,
-      "learning_rate": 4.545454545454545e-07,
+      "learning_rate": 2.2727272727272728e-06,
       "loss": 2.3833,
       "step": 1
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.07,
       "learning_rate": 9.090909090909091e-06,
-      "loss": 2.345,
-      "step": 20
+      "loss": 2.4762,
+      "step": 4
     },
     {
-      "epoch": 0.73,
+      "epoch": 0.15,
       "learning_rate": 1.8181818181818182e-05,
-      "loss": 2.333,
-      "step": 40
-    },
-    {
-      "epoch": 1.0,
-      "gpt4_scores": 0.21666666666666667,
-      "step": 55
+      "loss": 2.2934,
+      "step": 8
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 2.0215015411376953,
-      "eval_runtime": 4.9229,
-      "eval_samples_per_second": 4.672,
-      "eval_steps_per_second": 1.219,
-      "step": 55
-    },
-    {
-      "epoch": 1.09,
+      "epoch": 0.22,
       "learning_rate": 2.7272727272727273e-05,
-      "loss": 2.1124,
-      "step": 60
+      "loss": 2.2509,
+      "step": 12
     },
     {
-      "epoch": 1.45,
+      "epoch": 0.29,
       "learning_rate": 3.6363636363636364e-05,
-      "loss": 1.9167,
-      "step": 80
+      "loss": 2.1881,
+      "step": 16
     },
     {
-      "epoch": 1.82,
+      "epoch": 0.36,
       "learning_rate": 4.545454545454546e-05,
-      "loss": 1.8575,
-      "step": 100
-    },
-    {
-      "epoch": 2.0,
-      "gpt4_scores": 0.7166666666666667,
-      "step": 110
-    },
-    {
-      "epoch": 2.0,
-      "eval_loss": 1.8371087312698364,
-      "eval_runtime": 4.9624,
-      "eval_samples_per_second": 4.635,
-      "eval_steps_per_second": 1.209,
-      "step": 110
+      "loss": 2.2269,
+      "step": 20
     },
     {
-      "epoch": 2.18,
+      "epoch": 0.44,
       "learning_rate": 4.9987413559579636e-05,
-      "loss": 1.7124,
-      "step": 120
+      "loss": 2.1204,
+      "step": 24
     },
     {
-      "epoch": 2.55,
+      "epoch": 0.51,
       "learning_rate": 4.988679806432712e-05,
-      "loss": 1.7283,
-      "step": 140
+      "loss": 2.1803,
+      "step": 28
     },
     {
-      "epoch": 2.91,
+      "epoch": 0.58,
       "learning_rate": 4.968597221690986e-05,
-      "loss": 1.6435,
-      "step": 160
-    },
-    {
-      "epoch": 3.0,
-      "gpt4_scores": 0.7166666666666667,
-      "step": 165
+      "loss": 2.0278,
+      "step": 32
     },
     {
-      "epoch": 3.0,
-      "eval_loss": 1.8182122707366943,
-      "eval_runtime": 4.9633,
-      "eval_samples_per_second": 4.634,
-      "eval_steps_per_second": 1.209,
-      "step": 165
-    },
-    {
-      "epoch": 3.27,
+      "epoch": 0.65,
       "learning_rate": 4.938574467213518e-05,
-      "loss": 1.4997,
-      "step": 180
+      "loss": 1.9672,
+      "step": 36
     },
     {
-      "epoch": 3.64,
+      "epoch": 0.73,
       "learning_rate": 4.898732434036244e-05,
-      "loss": 1.4507,
-      "step": 200
+      "loss": 1.8425,
+      "step": 40
     },
     {
-      "epoch": 4.0,
+      "epoch": 0.8,
       "learning_rate": 4.849231551964771e-05,
-      "loss": 1.4138,
-      "step": 220
-    },
-    {
-      "epoch": 4.0,
-      "gpt4_scores": 0.7999999999999999,
-      "step": 220
-    },
-    {
-      "epoch": 4.0,
-      "eval_loss": 1.9195586442947388,
-      "eval_runtime": 4.9623,
-      "eval_samples_per_second": 4.635,
-      "eval_steps_per_second": 1.209,
-      "step": 220
+      "loss": 1.9323,
+      "step": 44
     },
     {
-      "epoch": 4.36,
+      "epoch": 0.87,
       "learning_rate": 4.790271143580174e-05,
-      "loss": 1.0929,
-      "step": 240
+      "loss": 2.0277,
+      "step": 48
     },
     {
-      "epoch": 4.73,
+      "epoch": 0.95,
       "learning_rate": 4.722088621637309e-05,
-      "loss": 1.0749,
-      "step": 260
+      "loss": 1.9256,
+      "step": 52
     },
     {
-      "epoch": 5.0,
-      "gpt4_scores": 0.6333333333333333,
-      "step": 275
+      "epoch": 1.0,
+      "gpt4_scores": 0.6833333333333332,
+      "step": 55
     },
     {
-      "epoch": 5.0,
-      "eval_loss": 2.2265050411224365,
-      "eval_runtime": 4.9358,
-      "eval_samples_per_second": 4.66,
-      "eval_steps_per_second": 1.216,
-      "step": 275
+      "epoch": 1.0,
+      "eval_loss": 1.878501534461975,
+      "eval_runtime": 4.9427,
+      "eval_samples_per_second": 4.653,
+      "eval_steps_per_second": 1.214,
+      "step": 55
     },
     {
-      "epoch": 5.09,
+      "epoch": 1.02,
       "learning_rate": 4.644958533087443e-05,
-      "loss": 0.9318,
-      "step": 280
+      "loss": 1.7957,
+      "step": 56
     },
     {
-      "epoch": 5.45,
+      "epoch": 1.09,
       "learning_rate": 4.559191453574582e-05,
-      "loss": 0.6893,
-      "step": 300
+      "loss": 1.8135,
+      "step": 60
     },
     {
-      "epoch": 5.82,
+      "epoch": 1.16,
       "learning_rate": 4.465132736856969e-05,
-      "loss": 0.6715,
-      "step": 320
-    },
-    {
-      "epoch": 6.0,
-      "gpt4_scores": 0.7666666666666666,
-      "step": 330
-    },
-    {
-      "epoch": 6.0,
-      "eval_loss": 2.5187041759490967,
-      "eval_runtime": 4.9618,
-      "eval_samples_per_second": 4.635,
-      "eval_steps_per_second": 1.209,
-      "step": 330
+      "loss": 1.7996,
+      "step": 64
     },
     {
-      "epoch": 6.18,
+      "epoch": 1.24,
       "learning_rate": 4.3631611241893874e-05,
-      "loss": 0.5555,
-      "step": 340
+      "loss": 1.7751,
+      "step": 68
     },
     {
-      "epoch": 6.55,
+      "epoch": 1.31,
       "learning_rate": 4.2536872192658036e-05,
-      "loss": 0.4251,
-      "step": 360
+      "loss": 1.8436,
+      "step": 72
     },
     {
-      "epoch": 6.91,
+      "epoch": 1.38,
       "learning_rate": 4.137151834863213e-05,
-      "loss": 0.4252,
-      "step": 380
-    },
-    {
-      "epoch": 7.0,
-      "gpt4_scores": 0.7833333333333333,
-      "step": 385
-    },
-    {
-      "epoch": 7.0,
-      "eval_loss": 2.830350875854492,
-      "eval_runtime": 4.9687,
-      "eval_samples_per_second": 4.629,
-      "eval_steps_per_second": 1.208,
-      "step": 385
+      "loss": 1.7179,
+      "step": 76
     },
     {
-      "epoch": 7.27,
+      "epoch": 1.45,
       "learning_rate": 4.014024217844167e-05,
-      "loss": 0.295,
-      "step": 400
+      "loss": 1.721,
+      "step": 80
     },
     {
-      "epoch": 7.64,
+      "epoch": 1.53,
       "learning_rate": 3.884800159665276e-05,
-      "loss": 0.2544,
-      "step": 420
+      "loss": 1.8576,
+      "step": 84
     },
     {
-      "epoch": 8.0,
+      "epoch": 1.6,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.2572,
-      "step": 440
-    },
-    {
-      "epoch": 8.0,
-      "gpt4_scores": 0.5833333333333334,
-      "step": 440
-    },
-    {
-      "epoch": 8.0,
-      "eval_loss": 3.170214891433716,
-      "eval_runtime": 4.9207,
-      "eval_samples_per_second": 4.674,
-      "eval_steps_per_second": 1.219,
-      "step": 440
+      "loss": 1.8062,
+      "step": 88
     },
     {
-      "epoch": 8.36,
+      "epoch": 1.67,
       "learning_rate": 3.610166531514436e-05,
-      "loss": 0.1568,
-      "step": 460
+      "loss": 1.6772,
+      "step": 92
     },
     {
-      "epoch": 8.73,
+      "epoch": 1.75,
       "learning_rate": 3.465862814232822e-05,
-      "loss": 0.1543,
-      "step": 480
-    },
-    {
-      "epoch": 9.0,
-      "gpt4_scores": 0.5833333333333334,
-      "step": 495
-    },
-    {
-      "epoch": 9.0,
-      "eval_loss": 3.3739395141601562,
-      "eval_runtime": 4.953,
-      "eval_samples_per_second": 4.644,
-      "eval_steps_per_second": 1.211,
-      "step": 495
+      "loss": 1.7685,
+      "step": 96
     },
     {
-      "epoch": 9.09,
+      "epoch": 1.82,
       "learning_rate": 3.3176699082935545e-05,
-      "loss": 0.1501,
-      "step": 500
+      "loss": 1.721,
+      "step": 100
     },
     {
-      "epoch": 9.45,
+      "epoch": 1.89,
       "learning_rate": 3.166184534225087e-05,
-      "loss": 0.1172,
-      "step": 520
+      "loss": 1.6726,
+      "step": 104
     },
     {
-      "epoch": 9.82,
+      "epoch": 1.96,
       "learning_rate": 3.012016670162977e-05,
-      "loss": 0.1095,
-      "step": 540
+      "loss": 1.6738,
+      "step": 108
     },
     {
-      "epoch": 10.0,
-      "gpt4_scores": 0.75,
-      "step": 550
+      "epoch": 2.0,
+      "gpt4_scores": 0.6666666666666666,
+      "step": 110
     },
     {
-      "epoch": 10.0,
-      "eval_loss": 3.496412515640259,
-      "eval_runtime": 4.961,
-      "eval_samples_per_second": 4.636,
-      "eval_steps_per_second": 1.209,
-      "step": 550
+      "epoch": 2.0,
+      "eval_loss": 1.8228907585144043,
+      "eval_runtime": 4.9969,
+      "eval_samples_per_second": 4.603,
+      "eval_steps_per_second": 1.201,
+      "step": 110
     },
     {
-      "epoch": 10.18,
+      "epoch": 2.04,
       "learning_rate": 2.8557870956832132e-05,
-      "loss": 0.1012,
-      "step": 560
+      "loss": 1.6321,
+      "step": 112
     },
     {
-      "epoch": 10.55,
+      "epoch": 2.11,
       "learning_rate": 2.698124892141971e-05,
-      "loss": 0.0825,
-      "step": 580
+      "loss": 1.6618,
+      "step": 116
     },
     {
-      "epoch": 10.91,
+      "epoch": 2.18,
       "learning_rate": 2.5396649095870202e-05,
-      "loss": 0.0875,
-      "step": 600
-    },
-    {
-      "epoch": 11.0,
-      "gpt4_scores": 0.48333333333333334,
-      "step": 605
-    },
-    {
-      "epoch": 11.0,
-      "eval_loss": 3.64684796333313,
-      "eval_runtime": 4.9227,
-      "eval_samples_per_second": 4.672,
-      "eval_steps_per_second": 1.219,
-      "step": 605
+      "loss": 1.5755,
+      "step": 120
     },
     {
-      "epoch": 11.27,
+      "epoch": 2.25,
       "learning_rate": 2.3810452104406444e-05,
-      "loss": 0.0739,
-      "step": 620
+      "loss": 1.6999,
+      "step": 124
     },
     {
-      "epoch": 11.64,
+      "epoch": 2.33,
       "learning_rate": 2.222904500247473e-05,
-      "loss": 0.0683,
-      "step": 640
+      "loss": 1.6391,
+      "step": 128
     },
     {
-      "epoch": 12.0,
+      "epoch": 2.4,
       "learning_rate": 2.0658795558326743e-05,
-      "loss": 0.0692,
-      "step": 660
-    },
-    {
-      "epoch": 12.0,
-      "gpt4_scores": 0.7166666666666667,
-      "step": 660
+      "loss": 1.583,
+      "step": 132
     },
     {
-      "epoch": 12.0,
-      "eval_loss": 3.800189256668091,
-      "eval_runtime": 4.9225,
-      "eval_samples_per_second": 4.672,
-      "eval_steps_per_second": 1.219,
-      "step": 660
-    },
-    {
-      "epoch": 12.36,
+      "epoch": 2.47,
       "learning_rate": 1.9106026612264316e-05,
-      "loss": 0.0617,
-      "step": 680
+      "loss": 1.6833,
+      "step": 136
     },
     {
-      "epoch": 12.73,
+      "epoch": 2.55,
       "learning_rate": 1.7576990616793137e-05,
-      "loss": 0.0573,
-      "step": 700
-    },
-    {
-      "epoch": 13.0,
-      "gpt4_scores": 0.5,
-      "step": 715
-    },
-    {
-      "epoch": 13.0,
-      "eval_loss": 3.918210506439209,
-      "eval_runtime": 4.9555,
-      "eval_samples_per_second": 4.641,
-      "eval_steps_per_second": 1.211,
-      "step": 715
+      "loss": 1.6736,
+      "step": 140
     },
     {
-      "epoch": 13.09,
+      "epoch": 2.62,
       "learning_rate": 1.6077844460203206e-05,
-      "loss": 0.058,
-      "step": 720
+      "loss": 1.5683,
+      "step": 144
     },
     {
-      "epoch": 13.45,
+      "epoch": 2.69,
       "learning_rate": 1.4614624674952842e-05,
-      "loss": 0.0512,
-      "step": 740
+      "loss": 1.5429,
+      "step": 148
     },
     {
-      "epoch": 13.82,
+      "epoch": 2.76,
       "learning_rate": 1.3193223130682936e-05,
-      "loss": 0.0527,
-      "step": 760
-    },
-    {
-      "epoch": 14.0,
-      "gpt4_scores": 0.5166666666666667,
-      "step": 770
+      "loss": 1.6129,
+      "step": 152
     },
     {
-      "epoch": 14.0,
-      "eval_loss": 4.011707782745361,
-      "eval_runtime": 4.9674,
-      "eval_samples_per_second": 4.63,
-      "eval_steps_per_second": 1.208,
-      "step": 770
-    },
-    {
-      "epoch": 14.18,
+      "epoch": 2.84,
       "learning_rate": 1.181936330973744e-05,
-      "loss": 0.0498,
-      "step": 780
+      "loss": 1.6105,
+      "step": 156
     },
     {
-      "epoch": 14.55,
+      "epoch": 2.91,
       "learning_rate": 1.049857726072005e-05,
-      "loss": 0.0489,
-      "step": 800
+      "loss": 1.5663,
+      "step": 160
     },
     {
-      "epoch": 14.91,
+      "epoch": 2.98,
       "learning_rate": 9.236183322886945e-06,
-      "loss": 0.0494,
-      "step": 820
+      "loss": 1.6181,
+      "step": 164
     },
     {
-      "epoch": 15.0,
-      "gpt4_scores": 0.65,
-      "step": 825
+      "epoch": 3.0,
+      "gpt4_scores": 0.7000000000000001,
+      "step": 165
     },
     {
-      "epoch": 15.0,
-      "eval_loss": 4.091977596282959,
-      "eval_runtime": 4.9428,
-      "eval_samples_per_second": 4.653,
-      "eval_steps_per_second": 1.214,
-      "step": 825
+      "epoch": 3.0,
+      "eval_loss": 1.8283250331878662,
+      "eval_runtime": 4.9725,
+      "eval_samples_per_second": 4.625,
+      "eval_steps_per_second": 1.207,
+      "step": 165
     },
     {
-      "epoch": 15.27,
+      "epoch": 3.05,
       "learning_rate": 8.0372647110717e-06,
-      "loss": 0.0471,
-      "step": 840
+      "loss": 1.5877,
+      "step": 168
     },
     {
-      "epoch": 15.64,
+      "epoch": 3.13,
       "learning_rate": 6.906649047373246e-06,
-      "loss": 0.0461,
-      "step": 860
+      "loss": 1.3838,
+      "step": 172
     },
     {
-      "epoch": 16.0,
+      "epoch": 3.2,
       "learning_rate": 5.848888922025553e-06,
-      "loss": 0.0459,
-      "step": 880
-    },
-    {
-      "epoch": 16.0,
-      "gpt4_scores": 0.5833333333333334,
-      "step": 880
+      "loss": 1.5406,
+      "step": 176
     },
     {
-      "epoch": 16.0,
-      "eval_loss": 4.127155303955078,
-      "eval_runtime": 4.9427,
-      "eval_samples_per_second": 4.653,
-      "eval_steps_per_second": 1.214,
-      "step": 880
-    },
-    {
-      "epoch": 16.36,
+      "epoch": 3.27,
       "learning_rate": 4.868243561723535e-06,
-      "loss": 0.0453,
-      "step": 900
+      "loss": 1.497,
+      "step": 180
     },
     {
-      "epoch": 16.73,
+      "epoch": 3.35,
       "learning_rate": 3.968661679220468e-06,
-      "loss": 0.0456,
-      "step": 920
-    },
-    {
-      "epoch": 17.0,
-      "gpt4_scores": 0.43333333333333335,
-      "step": 935
-    },
-    {
-      "epoch": 17.0,
-      "eval_loss": 4.1514201164245605,
-      "eval_runtime": 4.9682,
-      "eval_samples_per_second": 4.629,
-      "eval_steps_per_second": 1.208,
-      "step": 935
+      "loss": 1.5702,
+      "step": 184
     },
     {
-      "epoch": 17.09,
+      "epoch": 3.42,
       "learning_rate": 3.1537655732553768e-06,
-      "loss": 0.0469,
-      "step": 940
+      "loss": 1.6095,
+      "step": 188
     },
     {
-      "epoch": 17.45,
+      "epoch": 3.49,
       "learning_rate": 2.4268365428344736e-06,
-      "loss": 0.0447,
-      "step": 960
+      "loss": 1.4712,
+      "step": 192
     },
     {
-      "epoch": 17.82,
+      "epoch": 3.56,
       "learning_rate": 1.790801674598186e-06,
-      "loss": 0.0435,
-      "step": 980
+      "loss": 1.5552,
+      "step": 196
     },
     {
-      "epoch": 18.0,
-      "gpt4_scores": 0.6333333333333333,
-      "step": 990
-    },
-    {
-      "epoch": 18.0,
-      "eval_loss": 4.1617751121521,
-      "eval_runtime": 4.9681,
-      "eval_samples_per_second": 4.63,
-      "eval_steps_per_second": 1.208,
-      "step": 990
-    },
-    {
-      "epoch": 18.18,
+      "epoch": 3.64,
       "learning_rate": 1.248222056476367e-06,
-      "loss": 0.0484,
-      "step": 1000
+      "loss": 1.5805,
+      "step": 200
     },
     {
-      "epoch": 18.55,
+      "epoch": 3.71,
       "learning_rate": 8.012824650910938e-07,
-      "loss": 0.0436,
-      "step": 1020
+      "loss": 1.5588,
+      "step": 204
     },
     {
-      "epoch": 18.91,
+      "epoch": 3.78,
       "learning_rate": 4.517825684323324e-07,
-      "loss": 0.0446,
-      "step": 1040
-    },
-    {
-      "epoch": 19.0,
-      "gpt4_scores": 0.6166666666666667,
-      "step": 1045
-    },
-    {
-      "epoch": 19.0,
-      "eval_loss": 4.1660075187683105,
-      "eval_runtime": 4.9125,
-      "eval_samples_per_second": 4.682,
-      "eval_steps_per_second": 1.221,
-      "step": 1045
+      "loss": 1.64,
+      "step": 208
     },
     {
-      "epoch": 19.27,
+      "epoch": 3.85,
       "learning_rate": 2.011296792301165e-07,
-      "loss": 0.0463,
-      "step": 1060
+      "loss": 1.5666,
+      "step": 212
     },
     {
-      "epoch": 19.64,
+      "epoch": 3.93,
       "learning_rate": 5.033308820289184e-08,
-      "loss": 0.0473,
-      "step": 1080
+      "loss": 1.4496,
+      "step": 216
     },
     {
-      "epoch": 20.0,
+      "epoch": 4.0,
       "learning_rate": 0.0,
-      "loss": 0.0428,
-      "step": 1100
+      "loss": 1.5007,
+      "step": 220
     },
     {
-      "epoch": 20.0,
-      "gpt4_scores": 0.6833333333333332,
-      "step": 1100
+      "epoch": 4.0,
+      "gpt4_scores": 0.6666666666666666,
+      "step": 220
     },
     {
-      "epoch": 20.0,
-      "eval_loss": 4.166376113891602,
-      "eval_runtime": 4.9632,
-      "eval_samples_per_second": 4.634,
-      "eval_steps_per_second": 1.209,
-      "step": 1100
+      "epoch": 4.0,
+      "eval_loss": 1.8388779163360596,
+      "eval_runtime": 4.9378,
+      "eval_samples_per_second": 4.658,
+      "eval_steps_per_second": 1.215,
+      "step": 220
     },
     {
-      "epoch": 20.0,
-      "step": 1100,
-      "total_flos": 3.792163606268314e+16,
-      "train_loss": 0.5259156136621128,
-      "train_runtime": 7800.8143,
-      "train_samples_per_second": 0.556,
-      "train_steps_per_second": 0.141
+      "epoch": 4.0,
+      "step": 220,
+      "total_flos": 7586259830267904.0,
+      "train_loss": 1.7501471573656255,
+      "train_runtime": 1644.9971,
+      "train_samples_per_second": 0.528,
+      "train_steps_per_second": 0.134
     }
   ],
-  "logging_steps": 20,
-  "max_steps": 1100,
+  "logging_steps": 4,
+  "max_steps": 220,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 20,
+  "num_train_epochs": 4,
   "save_steps": 20,
-  "total_flos": 3.792163606268314e+16,
+  "total_flos": 7586259830267904.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null