End of training

README.md

@@ -1,13 +1,10 @@
 ---
-library_name: transformers
 base_model: EleutherAI/pythia-14m
 tags:
 - generated_from_trainer
 model-index:
 - name: pythia_14m_sft
   results: []
-datasets:
-- tatsu-lab/alpaca_farm
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -17,7 +14,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [EleutherAI/pythia-14m](https://huggingface.co/EleutherAI/pythia-14m) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 3.0709
+- Loss: 2.5974
 
 ## Model description
 
@@ -36,57 +33,26 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 2e-05
-- train_batch_size: 8
+- learning_rate: 5e-05
+- train_batch_size: 4
 - eval_batch_size: 8
 - seed: 42
-- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - num_epochs: 3
-- mixed_precision_training: Native AMP
 
 ### Training results
 
-| Training Loss | Epoch  | Step | Validation Loss |
-|:-------------:|:------:|:----:|:---------------:|
-| 3.5962        | 0.0889 | 100  | 3.4716          |
-| 3.4759        | 0.1778 | 200  | 3.4017          |
-| 3.4434        | 0.2667 | 300  | 3.3713          |
-| 3.3847        | 0.3556 | 400  | 3.3427          |
-| 3.359         | 0.4444 | 500  | 3.3186          |
-| 3.3447        | 0.5333 | 600  | 3.2899          |
-| 3.3099        | 0.6222 | 700  | 3.2654          |
-| 3.2775        | 0.7111 | 800  | 3.2482          |
-| 3.2512        | 0.8    | 900  | 3.2254          |
-| 3.2302        | 0.8889 | 1000 | 3.2151          |
-| 3.1752        | 0.9778 | 1100 | 3.1929          |
-| 3.1005        | 1.0667 | 1200 | 3.1916          |
-| 3.0798        | 1.1556 | 1300 | 3.1778          |
-| 3.073         | 1.2444 | 1400 | 3.1688          |
-| 3.1032        | 1.3333 | 1500 | 3.1707          |
-| 3.0417        | 1.4222 | 1600 | 3.1607          |
-| 3.075         | 1.5111 | 1700 | 3.1573          |
-| 3.0675        | 1.6    | 1800 | 3.1476          |
-| 3.0719        | 1.6889 | 1900 | 3.1424          |
-| 3.0469        | 1.7778 | 2000 | 3.1344          |
-| 3.0643        | 1.8667 | 2100 | 3.1257          |
-| 3.0375        | 1.9556 | 2200 | 3.1195          |
-| 2.9736        | 2.0444 | 2300 | 3.1130          |
-| 2.9915        | 2.1333 | 2400 | 3.1146          |
-| 2.9285        | 2.2222 | 2500 | 3.1061          |
-| 2.9535        | 2.3111 | 2600 | 3.0981          |
-| 2.9691        | 2.4    | 2700 | 3.0966          |
-| 2.9079        | 2.4889 | 2800 | 3.0889          |
-| 2.9216        | 2.5778 | 2900 | 3.0847          |
-| 2.924         | 2.6667 | 3000 | 3.0795          |
-| 2.9529        | 2.7556 | 3100 | 3.0775          |
-| 2.9375        | 2.8444 | 3200 | 3.0732          |
-| 2.8904        | 2.9333 | 3300 | 3.0709          |
+| Training Loss | Epoch | Step  | Validation Loss |
+|:-------------:|:-----:|:-----:|:---------------:|
+| 2.6472        | 1.0   | 4750  | 2.9652          |
+| 2.3754        | 2.0   | 9500  | 2.7470          |
+| 1.9951        | 3.0   | 14250 | 2.5974          |
 
 
 ### Framework versions
 
-- Transformers 4.46.2
-- Pytorch 2.5.1+cu121
-- Datasets 3.1.0
-- Tokenizers 0.20.3
+- Transformers 4.36.2
+- Pytorch 2.2.2+cu121
+- Datasets 2.18.0
+- Tokenizers 0.15.2

generation_config.json

@@ -2,5 +2,5 @@
   "_from_model_config": true,
   "bos_token_id": 0,
   "eos_token_id": 0,
-  "transformers_version": "4.46.2"
+  "transformers_version": "4.36.2"
 }