metadata
license: other
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
base_model: lightblue/suzume-llama-3-8B-japanese
datasets:
- generator
model-index:
- name: suzume-llama-3-8B-financial
results: []
suzume-llama-3-8B-financial
This model is a fine-tuned version of lightblue/suzume-llama-3-8B-japanese on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 1.6995
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: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_ratio: 0.03
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.2292 | 1.0 | 1 | 1.9473 |
| 0.8743 | 2.0 | 2 | 1.7062 |
| 0.633 | 3.0 | 3 | 1.6281 |
| 0.4712 | 4.0 | 4 | 1.6017 |
| 0.3204 | 5.0 | 5 | 1.6995 |
Framework versions
- PEFT 0.11.0
- Transformers 4.40.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1