|
--- |
|
license: other |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- HiTZ/alpaca_mt |
|
model-index: |
|
- name: alpaca-lora-30b-en-pt-es-ca-eu-gl-at |
|
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. --> |
|
|
|
# alpaca-lora-30b-en-pt-es-ca-eu-gl-at |
|
|
|
This model is a fine-tuned version of [decapoda-research/llama-30b-hf](https://huggingface.co/decapoda-research/llama-30b-hf) on the HiTZ/alpaca_mt ['en', 'pt', 'es', 'ca', 'eu', 'gl', 'at'] dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.9088 |
|
|
|
## 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: 6 |
|
- eval_batch_size: 6 |
|
- seed: 42 |
|
- distributed_type: multi-GPU |
|
- gradient_accumulation_steps: 21 |
|
- total_train_batch_size: 126 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_ratio: 0.03 |
|
- num_epochs: 1 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 1.1695 | 0.04 | 100 | 1.1716 | |
|
| 1.1211 | 0.07 | 200 | 1.0964 | |
|
| 1.0591 | 0.11 | 300 | 1.0590 | |
|
| 1.0234 | 0.14 | 400 | 1.0341 | |
|
| 1.0345 | 0.18 | 500 | 1.0165 | |
|
| 0.9932 | 0.22 | 600 | 1.0024 | |
|
| 0.9948 | 0.25 | 700 | 0.9895 | |
|
| 1.01 | 0.29 | 800 | 0.9794 | |
|
| 0.9488 | 0.32 | 900 | 0.9708 | |
|
| 0.9518 | 0.36 | 1000 | 0.9627 | |
|
| 0.9463 | 0.4 | 1100 | 0.9557 | |
|
| 0.956 | 0.43 | 1200 | 0.9498 | |
|
| 0.9521 | 0.47 | 1300 | 0.9437 | |
|
| 0.9345 | 0.51 | 1400 | 0.9385 | |
|
| 0.9469 | 0.54 | 1500 | 0.9337 | |
|
| 0.9466 | 0.58 | 1600 | 0.9297 | |
|
| 0.9403 | 0.61 | 1700 | 0.9257 | |
|
| 0.9179 | 0.65 | 1800 | 0.9219 | |
|
| 0.9468 | 0.69 | 1900 | 0.9190 | |
|
| 0.9173 | 0.72 | 2000 | 0.9163 | |
|
| 0.9172 | 0.76 | 2100 | 0.9142 | |
|
| 0.9351 | 0.79 | 2200 | 0.9124 | |
|
| 0.9238 | 0.83 | 2300 | 0.9110 | |
|
| 0.9057 | 0.87 | 2400 | 0.9099 | |
|
| 0.9309 | 0.9 | 2500 | 0.9093 | |
|
| 0.8893 | 0.94 | 2600 | 0.9090 | |
|
| 0.9095 | 0.97 | 2700 | 0.9088 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.28.0.dev0 |
|
- Pytorch 2.0.0+cu117 |
|
- Datasets 2.10.1 |
|
- Tokenizers 0.13.2 |
|
|
