Model save

README.md

@@ -20,7 +20,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [google/gemma-7b](https://huggingface.co/google/gemma-7b) on the generator dataset.
 It achieves the following results on the evaluation set:
-- Loss: 149.4047
+- Loss: 1.6581
 
 ## Model description
 
@@ -39,7 +39,7 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.03
+- learning_rate: 0.0002
 - train_batch_size: 8
 - eval_batch_size: 8
 - seed: 42
@@ -49,15 +49,15 @@ The following hyperparameters were used during training:
 - total_train_batch_size: 128
 - total_eval_batch_size: 64
 - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
-- lr_scheduler_type: linear
-- lr_scheduler_warmup_ratio: 0.06
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.1
 - num_epochs: 1
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 149.2635      | 1.0   | 140  | 149.4047        |
+| 1.6561        | 1.0   | 140  | 1.6581          |
 
 
 ### Framework versions

all_results.json

@@ -1,14 +1,9 @@
 {
     "epoch": 1.0,
-    "eval_loss": 149.4047088623047,
-    "eval_runtime": 2.8669,
-    "eval_samples": 518,
-    "eval_samples_per_second": 62.087,
-    "eval_steps_per_second": 1.046,
-    "total_flos": 4.2812236905630925e+17,
-    "train_loss": 249.82138957977295,
-    "train_runtime": 1603.7748,
+    "total_flos": 4.268849030789857e+17,
+    "train_loss": 5.955788305827549,
+    "train_runtime": 1733.7968,
     "train_samples": 51241,
-    "train_samples_per_second": 11.164,
-    "train_steps_per_second": 0.087
+    "train_samples_per_second": 10.326,
+    "train_steps_per_second": 0.081
 }

runs/Nov16_09-08-18_main-lora-gemma7b-alpaca-0-0/events.out.tfevents.1731766677.main-lora-gemma7b-alpaca-0-0.459.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6204c5fe75a8ba866469038f47376750537d637d555fdceb72f7df4437762ff1
-size 12689
+oid sha256:81d80ff5a57352be8e43e423954633ba553c054dca1139c35195a0bd83c46080
+size 13314

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 1.0,
-    "total_flos": 4.2812236905630925e+17,
-    "train_loss": 249.82138957977295,
-    "train_runtime": 1603.7748,
+    "total_flos": 4.268849030789857e+17,
+    "train_loss": 5.955788305827549,
+    "train_runtime": 1733.7968,
     "train_samples": 51241,
-    "train_samples_per_second": 11.164,
-    "train_steps_per_second": 0.087
+    "train_samples_per_second": 10.326,
+    "train_steps_per_second": 0.081
 }

trainer_state.json

@@ -10,223 +10,223 @@
   "log_history": [
     {
       "epoch": 0.007142857142857143,
-      "grad_norm": 286.327392578125,
-      "learning_rate": 0.003333333333333333,
+      "grad_norm": 106.10701751708984,
+      "learning_rate": 1.4285714285714285e-05,
       "loss": 27.4831,
       "step": 1
     },
     {
       "epoch": 0.03571428571428571,
-      "grad_norm": 564.7811889648438,
-      "learning_rate": 0.016666666666666666,
-      "loss": 180.9505,
+      "grad_norm": 46.149723052978516,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 26.2992,
       "step": 5
     },
     {
       "epoch": 0.07142857142857142,
-      "grad_norm": 45641.8671875,
-      "learning_rate": 0.029770992366412213,
-      "loss": 193.5063,
+      "grad_norm": 16.458189010620117,
+      "learning_rate": 0.00014285714285714287,
+      "loss": 20.4929,
       "step": 10
     },
     {
       "epoch": 0.10714285714285714,
-      "grad_norm": 1771723.5,
-      "learning_rate": 0.02862595419847328,
-      "loss": 376.766,
+      "grad_norm": 9.764805793762207,
+      "learning_rate": 0.00019996891820008164,
+      "loss": 16.0513,
       "step": 15
     },
     {
       "epoch": 0.14285714285714285,
-      "grad_norm": 286214.5,
-      "learning_rate": 0.02748091603053435,
-      "loss": 569.9451,
+      "grad_norm": 3.3296561241149902,
+      "learning_rate": 0.00019888308262251285,
+      "loss": 13.4526,
       "step": 20
     },
     {
       "epoch": 0.17857142857142858,
-      "grad_norm": 1100.7747802734375,
-      "learning_rate": 0.02633587786259542,
-      "loss": 615.8839,
+      "grad_norm": 3.235478162765503,
+      "learning_rate": 0.0001962624246950012,
+      "loss": 12.6172,
       "step": 25
     },
     {
       "epoch": 0.21428571428571427,
-      "grad_norm": 169022.0,
-      "learning_rate": 0.025190839694656485,
-      "loss": 481.0353,
+      "grad_norm": 5.432589530944824,
+      "learning_rate": 0.00019214762118704076,
+      "loss": 11.9808,
       "step": 30
     },
     {
       "epoch": 0.25,
-      "grad_norm": 333.40972900390625,
-      "learning_rate": 0.024045801526717557,
-      "loss": 305.1498,
+      "grad_norm": 10.644349098205566,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 11.0065,
       "step": 35
     },
     {
       "epoch": 0.2857142857142857,
-      "grad_norm": 139948.15625,
-      "learning_rate": 0.022900763358778626,
-      "loss": 241.1332,
+      "grad_norm": 14.282907485961914,
+      "learning_rate": 0.00017971325072229226,
+      "loss": 9.3131,
       "step": 40
     },
     {
       "epoch": 0.32142857142857145,
-      "grad_norm": 1393742.875,
-      "learning_rate": 0.021755725190839695,
-      "loss": 207.3882,
+      "grad_norm": 21.60712432861328,
+      "learning_rate": 0.00017158668492597186,
+      "loss": 7.1723,
       "step": 45
     },
     {
       "epoch": 0.35714285714285715,
-      "grad_norm": 12597.32421875,
-      "learning_rate": 0.020610687022900764,
-      "loss": 201.0974,
+      "grad_norm": 15.418853759765625,
+      "learning_rate": 0.00016234898018587337,
+      "loss": 4.6599,
       "step": 50
     },
     {
       "epoch": 0.39285714285714285,
-      "grad_norm": 1081976.875,
-      "learning_rate": 0.01946564885496183,
-      "loss": 222.2186,
+      "grad_norm": 5.335650444030762,
+      "learning_rate": 0.0001521435203379498,
+      "loss": 2.855,
       "step": 55
     },
     {
       "epoch": 0.42857142857142855,
-      "grad_norm": 18104.56640625,
-      "learning_rate": 0.0183206106870229,
-      "loss": 254.4615,
+      "grad_norm": 4.913125038146973,
+      "learning_rate": 0.00014112871031306119,
+      "loss": 2.3938,
       "step": 60
     },
     {
       "epoch": 0.4642857142857143,
-      "grad_norm": 53967.2265625,
-      "learning_rate": 0.017175572519083967,
-      "loss": 256.7784,
+      "grad_norm": 3.951244592666626,
+      "learning_rate": 0.00012947551744109043,
+      "loss": 2.1646,
       "step": 65
     },
     {
       "epoch": 0.5,
-      "grad_norm": 36778.24609375,
-      "learning_rate": 0.01603053435114504,
-      "loss": 189.0749,
+      "grad_norm": 1.6933783292770386,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 2.0061,
       "step": 70
     },
     {
       "epoch": 0.5357142857142857,
-      "grad_norm": 1759.3746337890625,
-      "learning_rate": 0.014885496183206106,
-      "loss": 198.0302,
+      "grad_norm": 0.9344208836555481,
+      "learning_rate": 0.00010498458856606972,
+      "loss": 1.9092,
       "step": 75
     },
     {
       "epoch": 0.5714285714285714,
-      "grad_norm": 2888.434326171875,
-      "learning_rate": 0.013740458015267175,
-      "loss": 208.9426,
+      "grad_norm": 0.9437263011932373,
+      "learning_rate": 9.252699064135758e-05,
+      "loss": 1.8086,
       "step": 80
     },
     {
       "epoch": 0.6071428571428571,
-      "grad_norm": 3183.7255859375,
-      "learning_rate": 0.012595419847328242,
-      "loss": 237.6604,
+      "grad_norm": 1.5205684900283813,
+      "learning_rate": 8.018538568006027e-05,
+      "loss": 1.7838,
       "step": 85
     },
     {
       "epoch": 0.6428571428571429,
-      "grad_norm": 295.5591735839844,
-      "learning_rate": 0.011450381679389313,
-      "loss": 259.345,
+      "grad_norm": 0.8009554743766785,
+      "learning_rate": 6.815133497483157e-05,
+      "loss": 1.7433,
       "step": 90
     },
     {
       "epoch": 0.6785714285714286,
-      "grad_norm": 2817.36083984375,
-      "learning_rate": 0.010305343511450382,
-      "loss": 229.8393,
+      "grad_norm": 0.9393540620803833,
+      "learning_rate": 5.6611626088244194e-05,
+      "loss": 1.736,
       "step": 95
     },
     {
       "epoch": 0.7142857142857143,
-      "grad_norm": 1439.1002197265625,
-      "learning_rate": 0.00916030534351145,
-      "loss": 206.7085,
+      "grad_norm": 1.6479978561401367,
+      "learning_rate": 4.574537361342407e-05,
+      "loss": 1.6976,
       "step": 100
     },
     {
       "epoch": 0.75,
-      "grad_norm": 2063.854248046875,
-      "learning_rate": 0.00801526717557252,
-      "loss": 200.2123,
+      "grad_norm": 0.7829992771148682,
+      "learning_rate": 3.5721239031346066e-05,
+      "loss": 1.6793,
       "step": 105
     },
     {
       "epoch": 0.7857142857142857,
-      "grad_norm": 3193.01953125,
-      "learning_rate": 0.006870229007633588,
-      "loss": 195.9004,
+      "grad_norm": 0.7236832976341248,
+      "learning_rate": 2.669481281701739e-05,
+      "loss": 1.7,
       "step": 110
     },
     {
       "epoch": 0.8214285714285714,
-      "grad_norm": 1513.8978271484375,
-      "learning_rate": 0.0057251908396946565,
-      "loss": 188.8404,
+      "grad_norm": 0.7881686687469482,
+      "learning_rate": 1.880619942841435e-05,
+      "loss": 1.6778,
       "step": 115
     },
     {
       "epoch": 0.8571428571428571,
-      "grad_norm": 5677.81494140625,
-      "learning_rate": 0.004580152671755725,
-      "loss": 176.7594,
+      "grad_norm": 1.7755261659622192,
+      "learning_rate": 1.2177842662977135e-05,
+      "loss": 1.6757,
       "step": 120
     },
     {
       "epoch": 0.8928571428571429,
-      "grad_norm": 726.0751953125,
-      "learning_rate": 0.003435114503816794,
-      "loss": 167.6159,
+      "grad_norm": 1.0512515306472778,
+      "learning_rate": 6.9126251355795864e-06,
+      "loss": 1.6766,
       "step": 125
     },
     {
       "epoch": 0.9285714285714286,
-      "grad_norm": 1151.048095703125,
-      "learning_rate": 0.0022900763358778627,
-      "loss": 155.8075,
+      "grad_norm": 0.7070327997207642,
+      "learning_rate": 3.092271377092215e-06,
+      "loss": 1.67,
       "step": 130
     },
     {
       "epoch": 0.9642857142857143,
-      "grad_norm": 3488.1396484375,
-      "learning_rate": 0.0011450381679389313,
-      "loss": 155.3781,
+      "grad_norm": 0.7091158628463745,
+      "learning_rate": 7.760793399827937e-07,
+      "loss": 1.6459,
       "step": 135
     },
     {
       "epoch": 1.0,
-      "grad_norm": 718.7391357421875,
+      "grad_norm": 0.7374704480171204,
       "learning_rate": 0.0,
-      "loss": 149.2635,
+      "loss": 1.6561,
       "step": 140
     },
     {
       "epoch": 1.0,
-      "eval_loss": 149.4047088623047,
-      "eval_runtime": 2.8984,
-      "eval_samples_per_second": 61.413,
-      "eval_steps_per_second": 1.035,
+      "eval_loss": 1.6581236124038696,
+      "eval_runtime": 3.8242,
+      "eval_samples_per_second": 46.545,
+      "eval_steps_per_second": 0.784,
       "step": 140
     },
     {
       "epoch": 1.0,
       "step": 140,
-      "total_flos": 4.2812236905630925e+17,
-      "train_loss": 249.82138957977295,
-      "train_runtime": 1603.7748,
-      "train_samples_per_second": 11.164,
-      "train_steps_per_second": 0.087
+      "total_flos": 4.268849030789857e+17,
+      "train_loss": 5.955788305827549,
+      "train_runtime": 1733.7968,
+      "train_samples_per_second": 10.326,
+      "train_steps_per_second": 0.081
     }
   ],
   "logging_steps": 5,
@@ -246,7 +246,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 4.2812236905630925e+17,
+  "total_flos": 4.268849030789857e+17,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null