Training in progress, epoch 1

README.md

@@ -1,24 +1,20 @@
 ---
-library_name: transformers
-license: apache-2.0
-base_model: EleutherAI/pythia-70m
+base_model: EleutherAI/pythia-14m
 tags:
 - generated_from_trainer
 model-index:
-- name: pythia_70m_sft
+- 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
 should probably proofread and complete it, then remove this comment. -->
 
-# pythia_70m_sft
+# pythia_14m_sft
 
-This model is a fine-tuned version of [EleutherAI/pythia-70m](https://huggingface.co/EleutherAI/pythia-70m) on an unknown dataset.
+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: 2.4023
+- Loss: 2.4367
 
 ## Model description
 
@@ -37,57 +33,26 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 2e-05
-- train_batch_size: 8
-- eval_batch_size: 8
+- learning_rate: 5e-05
+- train_batch_size: 32
+- eval_batch_size: 64
 - 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 |
-|:-------------:|:------:|:----:|:---------------:|
-| 2.8612        | 0.0889 | 100  | 2.7705          |
-| 2.7799        | 0.1778 | 200  | 2.7116          |
-| 2.754         | 0.2667 | 300  | 2.6827          |
-| 2.7035        | 0.3556 | 400  | 2.6482          |
-| 2.6667        | 0.4444 | 500  | 2.6296          |
-| 2.6568        | 0.5333 | 600  | 2.6048          |
-| 2.6233        | 0.6222 | 700  | 2.5862          |
-| 2.5956        | 0.7111 | 800  | 2.5790          |
-| 2.5635        | 0.8    | 900  | 2.5436          |
-| 2.5469        | 0.8889 | 1000 | 2.5445          |
-| 2.499         | 0.9778 | 1100 | 2.5155          |
-| 2.3677        | 1.0667 | 1200 | 2.5158          |
-| 2.326         | 1.1556 | 1300 | 2.5018          |
-| 2.3132        | 1.2444 | 1400 | 2.5023          |
-| 2.3401        | 1.3333 | 1500 | 2.5026          |
-| 2.3027        | 1.4222 | 1600 | 2.4953          |
-| 2.3223        | 1.5111 | 1700 | 2.4804          |
-| 2.3186        | 1.6    | 1800 | 2.4776          |
-| 2.3187        | 1.6889 | 1900 | 2.4709          |
-| 2.3102        | 1.7778 | 2000 | 2.4618          |
-| 2.3129        | 1.8667 | 2100 | 2.4526          |
-| 2.287         | 1.9556 | 2200 | 2.4456          |
-| 2.1946        | 2.0444 | 2300 | 2.4423          |
-| 2.1757        | 2.1333 | 2400 | 2.4408          |
-| 2.1308        | 2.2222 | 2500 | 2.4383          |
-| 2.1475        | 2.3111 | 2600 | 2.4283          |
-| 2.155         | 2.4    | 2700 | 2.4231          |
-| 2.1197        | 2.4889 | 2800 | 2.4222          |
-| 2.1225        | 2.5778 | 2900 | 2.4173          |
-| 2.1196        | 2.6667 | 3000 | 2.4118          |
-| 2.146         | 2.7556 | 3100 | 2.4075          |
-| 2.1361        | 2.8444 | 3200 | 2.4046          |
-| 2.0965        | 2.9333 | 3300 | 2.4023          |
+| Training Loss | Epoch | Step | Validation Loss |
+|:-------------:|:-----:|:----:|:---------------:|
+| 2.5074        | 1.0   | 594  | 2.5707          |
+| 2.2083        | 2.0   | 1188 | 2.4775          |
+| 2.1035        | 3.0   | 1782 | 2.4367          |
 
 
 ### 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

config.json

@@ -3,7 +3,6 @@
   "architectures": [
     "GPTNeoXForCausalLM"
   ],
-  "attention_bias": true,
   "attention_dropout": 0.0,
   "bos_token_id": 0,
   "classifier_dropout": 0.1,
@@ -18,14 +17,13 @@
   "model_type": "gpt_neox",
   "num_attention_heads": 8,
   "num_hidden_layers": 6,
-  "partial_rotary_factor": 0.25,
+  "pad_token_id": 0,
   "rope_scaling": null,
-  "rope_theta": 10000,
   "rotary_emb_base": 10000,
   "rotary_pct": 0.25,
   "tie_word_embeddings": false,
   "torch_dtype": "float32",
-  "transformers_version": "4.46.2",
+  "transformers_version": "4.36.2",
   "use_cache": true,
   "use_parallel_residual": true,
   "vocab_size": 50304

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"
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2d9ae519a3a705e4684e500763156440c7a05a6e123acfda46df48dcb57bf7b3
+oid sha256:f0b2fcd60669593560c940aaee06564d68290dbc1955894b96fe18ae795869e6
 size 281715176

tokenizer_config.json

@@ -1,6 +1,4 @@
 {
-  "add_bos_token": false,
-  "add_eos_token": false,
   "add_prefix_space": false,
   "added_tokens_decoder": {
     "0": {
@@ -205,7 +203,7 @@
     }
   },
   "bos_token": "<|endoftext|>",
-  "clean_up_tokenization_spaces": false,
+  "clean_up_tokenization_spaces": true,
   "eos_token": "<|endoftext|>",
   "model_max_length": 1000000000000000019884624838656,
   "pad_token": "<|endoftext|>",

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:372d5c4f9bc9b66ca290960c1ea9131b8232d5334e429529fd6a9adeab4d94cf
-size 5240
+oid sha256:4518f5df23b9750a1f447908b799bba58857aa55ee791669d840ab4c9223fb6c
+size 4664