|
--- |
|
license: gemma |
|
base_model: google/gemma-2b |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: GOLM3 |
|
results: [] |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# GOLM3 |
|
|
|
This model is a fine-tuned version of [google/gemma-2b](https://huggingface.co/google/gemma-2b) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.1016 |
|
|
|
## Model description |
|
|
|
More information needed |
|
|
|
## Intended uses & limitations |
|
|
|
More information needed |
|
|
|
## Training and evaluation data |
|
|
|
More information needed |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 0.0003 |
|
- train_batch_size: 8 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 16 |
|
- total_train_batch_size: 128 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine_with_restarts |
|
- lr_scheduler_warmup_steps: 80 |
|
- num_epochs: 3 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 1.7996 | 0.09 | 10 | 1.4084 | |
|
| 0.9949 | 0.18 | 20 | 0.5027 | |
|
| 0.3011 | 0.27 | 30 | 0.1578 | |
|
| 0.1527 | 0.36 | 40 | 0.1481 | |
|
| 0.1447 | 0.45 | 50 | 0.1469 | |
|
| 0.1451 | 0.54 | 60 | 0.1464 | |
|
| 0.142 | 0.63 | 70 | 0.1422 | |
|
| 0.1422 | 0.73 | 80 | 0.1372 | |
|
| 0.1304 | 0.82 | 90 | 0.1289 | |
|
| 0.1241 | 0.91 | 100 | 0.1269 | |
|
| 0.1263 | 1.0 | 110 | 0.1302 | |
|
| 0.1163 | 1.09 | 120 | 0.1185 | |
|
| 0.1091 | 1.18 | 130 | 0.1211 | |
|
| 0.1143 | 1.27 | 140 | 0.1143 | |
|
| 0.1131 | 1.36 | 150 | 0.1113 | |
|
| 0.1127 | 1.45 | 160 | 0.1115 | |
|
| 0.1087 | 1.54 | 170 | 0.1073 | |
|
| 0.1086 | 1.63 | 180 | 0.1064 | |
|
| 0.1069 | 1.72 | 190 | 0.1053 | |
|
| 0.1027 | 1.81 | 200 | 0.1047 | |
|
| 0.1037 | 1.9 | 210 | 0.1022 | |
|
| 0.1072 | 1.99 | 220 | 0.1029 | |
|
| 0.0896 | 2.08 | 230 | 0.1056 | |
|
| 0.0918 | 2.18 | 240 | 0.1024 | |
|
| 0.0828 | 2.27 | 250 | 0.1026 | |
|
| 0.0861 | 2.36 | 260 | 0.1022 | |
|
| 0.0853 | 2.45 | 270 | 0.1049 | |
|
| 0.0809 | 2.54 | 280 | 0.1028 | |
|
| 0.0782 | 2.63 | 290 | 0.1021 | |
|
| 0.0814 | 2.72 | 300 | 0.1021 | |
|
| 0.0849 | 2.81 | 310 | 0.1019 | |
|
| 0.0838 | 2.9 | 320 | 0.1016 | |
|
| 0.0864 | 2.99 | 330 | 0.1016 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.36.0.dev0 |
|
- Pytorch 2.1.2+cu121 |
|
- Datasets 2.14.6 |
|
- Tokenizers 0.14.1 |
|
|
