metadata
library_name: peft
license: other
base_model: /root/autodl-fs/Qwen2.5-7B-Instruct
tags:
- llama-factory
- lora
- generated_from_trainer
model-index:
- name: stage1_cap_func
results: []
stage1_cap_func
This model is a fine-tuned version of /root/autodl-fs/Qwen2.5-7B-Instruct on the textui-caption2k and the textui-function6k datasets. It achieves the following results on the evaluation set:
- Loss: 0.4610
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: 0.0001
- train_batch_size: 3
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 8
- total_train_batch_size: 96
- total_eval_batch_size: 16
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 25
- training_steps: 250
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.6383 | 0.2436 | 25 | 0.6979 |
| 0.5722 | 0.4872 | 50 | 0.6059 |
| 0.5268 | 0.7308 | 75 | 0.5485 |
| 0.492 | 0.9744 | 100 | 0.5159 |
| 0.4719 | 1.2241 | 125 | 0.4990 |
| 0.4031 | 1.4677 | 150 | 0.4809 |
| 0.3825 | 1.7113 | 175 | 0.4702 |
| 0.3944 | 1.9549 | 200 | 0.4625 |
| 0.3411 | 2.2046 | 225 | 0.4616 |
| 0.3588 | 2.4482 | 250 | 0.4610 |
Framework versions
- PEFT 0.15.1
- Transformers 4.51.3
- Pytorch 2.7.0+cu126
- Datasets 3.5.0
- Tokenizers 0.21.1