README.md

---
base_model: deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
library_name: transformers
model_name: DeepSeek-R1-Distill-Qwen-1.5B-Medical-QLoRA
tags:
- generated_from_trainer
- trl
- sft
licence: license
license: mit
datasets:
- FreedomIntelligence/medical-o1-reasoning-SFT
language:
- en
pipeline_tag: text-generation
---

# DeepSeek-R1-Distill-Qwen-1.5B-Medical-QLoRA

This is an adapter of [MilyaShams/DeepSeek-R1-Distill-Qwen-1.5B-Medical](https://huggingface.co/MilyaShams/DeepSeek-R1-Distill-Qwen-1.5B-Medical), fine-tuned using [QLoRA](https://huggingface.co/docs/peft/main/en/conceptual_guides/lora) with [TRL](https://github.com/huggingface/trl).

The adapter preserves the medical domain specialization of the base model while optimizing memory efficiency and training speed through low-rank adaptation.


## Quick start

```python
from transformers import pipeline

question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="MilyaShams/DeepSeek-R1-Distill-Qwen-1.5B-Medical-QLoRA", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=1024, return_full_text=False)[0]
print(output["generated_text"])
```

## Training procedure

[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/miliusha2801-innopolis-university/Deepseek-R1-Qwen-1.5b%20SFT%20on%20medical%20dataset%20full%201%20epoch%20v.0/runs/7q51lr76) 


This model was trained with SFT.

### Framework versions

- TRL: 0.15.2
- Transformers: 4.47.0
- Pytorch: 2.5.1+cu121
- Datasets: 3.3.1
- Tokenizers: 0.21.0

## Citations



Cite TRL as:
    
```bibtex
@misc{vonwerra2022trl,
	title        = {{TRL: Transformer Reinforcement Learning}},
	author       = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
	year         = 2020,
	journal      = {GitHub repository},
	publisher    = {GitHub},
	howpublished = {\url{https://github.com/huggingface/trl}}
}
```