|
|
--- |
|
|
language: ko |
|
|
license: mit |
|
|
metrics: |
|
|
- perplexity |
|
|
- accuracy |
|
|
tags: |
|
|
- korean |
|
|
- qwen |
|
|
- finetunned |
|
|
dataset_tags: |
|
|
- kyujinpy/KOpen-platypus |
|
|
--- |
|
|
|
|
|
# Qwen 2.5 3B Instruction-tuned Model |
|
|
|
|
|
This model is a Instruction-tuned version of Qwen 2.5 3B for recipie recommandation. |
|
|
|
|
|
## Model Description |
|
|
|
|
|
- Fine-tuned from: Qwen/Qwen2.5-3B |
|
|
- Fine-tuning task: [Instruction-tuning] |
|
|
- Training data: [kyujinpy/KOpen-platypus + Recipe data] |
|
|
- Evaluation results: [Add your evaluation metrics] |
|
|
|
|
|
## Usage |
|
|
|
|
|
```python |
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
|
|
base_model_path = "Qwen/Qwen2.5-3B" |
|
|
adapter_model = "INo0121/qwen2.5_3b_instruction_tuning_241020" |
|
|
base_model = AutoModelForCausalLM.from_pretrained( |
|
|
base_model_path, |
|
|
torch_dtype="auto", |
|
|
device_map="auto", |
|
|
temperature=0.1 |
|
|
) |
|
|
model = PeftModel.from_pretrained(base_model, adapter_model) |
|
|
tokenizer = AutoTokenizer.from_pretrained(adapter_model) |
|
|
|
|
|
|
|
|
# Example usage |
|
|
input_text = "Your input text here" |
|
|
inputs = tokenizer(input_text, return_tensors="pt") |
|
|
outputs = model.generate(**inputs) |
|
|
print(tokenizer.decode(outputs[0], skip_special_tokens=True)) |
|
|
``` |
|
|
|
|
|
## Limitations and Biases |
|
|
|
|
|
[Describe any known limitations or biases of your model] |
|
|
|
|
|
## Training Details |
|
|
|
|
|
- Training framework: Hugging Face Transformers |
|
|
- Hyperparameters: [List your key hyperparameters] |
|
|
- Training hardware: [Describe the hardware used] |
|
|
|
