|
--- |
|
license: gemma |
|
base_model: google/gemma-2-9b |
|
tags: |
|
- trl |
|
- sft |
|
- generated_from_trainer |
|
model-index: |
|
- name: collapse_gemma-2-9b_hs2_accumulate_iter2_sftsd0 |
|
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. --> |
|
|
|
# collapse_gemma-2-9b_hs2_accumulate_iter2_sftsd0 |
|
|
|
This model is a fine-tuned version of [google/gemma-2-9b](https://huggingface.co/google/gemma-2-9b) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.9438 |
|
- Num Input Tokens Seen: 9944616 |
|
|
|
## 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: 8e-06 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 16 |
|
- seed: 0 |
|
- gradient_accumulation_steps: 32 |
|
- total_train_batch_size: 128 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: constant_with_warmup |
|
- lr_scheduler_warmup_ratio: 0.05 |
|
- num_epochs: 1 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Input Tokens Seen | |
|
|:-------------:|:------:|:----:|:---------------:|:-----------------:| |
|
| No log | 0 | 0 | 1.2335 | 0 | |
|
| 1.1198 | 0.0263 | 5 | 1.1072 | 256304 | |
|
| 1.0654 | 0.0527 | 10 | 1.0185 | 519116 | |
|
| 0.8862 | 0.0790 | 15 | 0.9889 | 775168 | |
|
| 0.8666 | 0.1054 | 20 | 0.9891 | 1038920 | |
|
| 0.7782 | 0.1317 | 25 | 0.9886 | 1306500 | |
|
| 0.6537 | 0.1581 | 30 | 0.9872 | 1568200 | |
|
| 0.7345 | 0.1844 | 35 | 0.9877 | 1831700 | |
|
| 0.6292 | 0.2107 | 40 | 0.9795 | 2092712 | |
|
| 0.6696 | 0.2371 | 45 | 0.9755 | 2353476 | |
|
| 0.5445 | 0.2634 | 50 | 0.9722 | 2620524 | |
|
| 0.6364 | 0.2898 | 55 | 0.9687 | 2886160 | |
|
| 0.6564 | 0.3161 | 60 | 0.9671 | 3149304 | |
|
| 0.5167 | 0.3424 | 65 | 0.9640 | 3413380 | |
|
| 0.6553 | 0.3688 | 70 | 0.9627 | 3684636 | |
|
| 0.5201 | 0.3951 | 75 | 0.9603 | 3947600 | |
|
| 0.5839 | 0.4215 | 80 | 0.9603 | 4207528 | |
|
| 0.5599 | 0.4478 | 85 | 0.9587 | 4468996 | |
|
| 0.6981 | 0.4742 | 90 | 0.9590 | 4730728 | |
|
| 0.582 | 0.5005 | 95 | 0.9558 | 4991328 | |
|
| 0.5174 | 0.5268 | 100 | 0.9556 | 5253436 | |
|
| 0.6031 | 0.5532 | 105 | 0.9545 | 5518624 | |
|
| 0.6314 | 0.5795 | 110 | 0.9528 | 5780988 | |
|
| 0.4925 | 0.6059 | 115 | 0.9527 | 6041796 | |
|
| 0.5823 | 0.6322 | 120 | 0.9515 | 6307948 | |
|
| 0.5974 | 0.6585 | 125 | 0.9498 | 6573748 | |
|
| 0.4411 | 0.6849 | 130 | 0.9492 | 6836544 | |
|
| 0.4604 | 0.7112 | 135 | 0.9489 | 7098504 | |
|
| 0.564 | 0.7376 | 140 | 0.9475 | 7354740 | |
|
| 0.5769 | 0.7639 | 145 | 0.9477 | 7620140 | |
|
| 0.4886 | 0.7903 | 150 | 0.9468 | 7884420 | |
|
| 0.5637 | 0.8166 | 155 | 0.9462 | 8151036 | |
|
| 0.5161 | 0.8429 | 160 | 0.9460 | 8414540 | |
|
| 0.633 | 0.8693 | 165 | 0.9459 | 8677992 | |
|
| 0.5239 | 0.8956 | 170 | 0.9446 | 8937256 | |
|
| 0.6149 | 0.9220 | 175 | 0.9465 | 9204996 | |
|
| 0.5386 | 0.9483 | 180 | 0.9451 | 9467132 | |
|
| 0.6638 | 0.9746 | 185 | 0.9446 | 9732120 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.44.0 |
|
- Pytorch 2.4.0+cu121 |
|
- Datasets 2.20.0 |
|
- Tokenizers 0.19.1 |
|
|
