metadata
license: other
library_name: peft
tags:
- generated_from_trainer
base_model: astronomer/Llama-3-8B-Instruct-GPTQ-4-Bit
model-index:
- name: agrobot-llama3-0.2-ft
results: []
agrobot-llama3-0.2-ft
This model is a fine-tuned version of astronomer/Llama-3-8B-Instruct-GPTQ-4-Bit on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4271
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: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.0139 | 0.99 | 102 | 0.4580 |
| 0.4463 | 2.0 | 205 | 0.4415 |
| 0.4292 | 3.0 | 308 | 0.4318 |
| 0.4159 | 4.0 | 411 | 0.4266 |
| 0.4063 | 4.99 | 513 | 0.4232 |
| 0.3898 | 6.0 | 616 | 0.4222 |
| 0.3778 | 7.0 | 719 | 0.4232 |
| 0.3672 | 8.0 | 822 | 0.4240 |
| 0.3617 | 8.99 | 924 | 0.4264 |
| 0.3503 | 9.93 | 1020 | 0.4271 |
Framework versions
- PEFT 0.11.1
- Transformers 4.39.3
- Pytorch 2.1.0+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2