|
--- |
|
base_model: ahxt/LiteLlama-460M-1T |
|
library_name: peft |
|
license: mit |
|
tags: |
|
- trl |
|
- sft |
|
- generated_from_trainer |
|
model-index: |
|
- name: ColdLLamaLite |
|
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. --> |
|
|
|
# ColdLLamaLite |
|
|
|
This model is a fine-tuned version of [ahxt/LiteLlama-460M-1T](https://huggingface.co/ahxt/LiteLlama-460M-1T) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 2.0471 |
|
|
|
## 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: 0.0002 |
|
- train_batch_size: 32 |
|
- eval_batch_size: 32 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 256 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_ratio: 0.03 |
|
- num_epochs: 10 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 4.1747 | 0.8 | 25 | 3.9257 | |
|
| 3.626 | 1.6 | 50 | 3.2474 | |
|
| 2.8441 | 2.4 | 75 | 2.4490 | |
|
| 2.3365 | 3.2 | 100 | 2.2482 | |
|
| 2.2153 | 4.0 | 125 | 2.1758 | |
|
| 2.1591 | 4.8 | 150 | 2.1316 | |
|
| 2.1214 | 5.6 | 175 | 2.1011 | |
|
| 2.0946 | 6.4 | 200 | 2.0781 | |
|
| 2.0818 | 7.2 | 225 | 2.0622 | |
|
| 2.0614 | 8.0 | 250 | 2.0528 | |
|
| 2.0571 | 8.8 | 275 | 2.0485 | |
|
| 2.0522 | 9.6 | 300 | 2.0471 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.12.0 |
|
- Transformers 4.43.3 |
|
- Pytorch 2.3.1+cu121 |
|
- Datasets 2.17.0 |
|
- Tokenizers 0.19.1 |
