|
|
--- |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- FreedomIntelligence/medical-o1-reasoning-SFT |
|
|
language: |
|
|
- zh |
|
|
base_model: |
|
|
- unsloth/DeepSeek-R1-Distill-Llama-8B |
|
|
pipeline_tag: text-generation |
|
|
tags: |
|
|
- unsloth |
|
|
- deepseek |
|
|
- rola |
|
|
- sft |
|
|
--- |
|
|
|
|
|
# DeepSeek-R1 Medical Reasoning Model |
|
|
|
|
|
This repository contains a **fine-tuned medical reasoning model** based on |
|
|
[DeepSeek-R1-Distill-Llama-8B](https://huggingface.co/unsloth/DeepSeek-R1-Distill-Llama-8B) |
|
|
and trained on the [medical-o1-reasoning-SFT](https://huggingface.co/datasets/FreedomIntelligence/medical-o1-reasoning-SFT) dataset. |
|
|
|
|
|
⚠️ **The uploaded file (`unsloth.Q8_0.gguf`) contains quantized weights** for efficient inference. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🔍 Model Overview |
|
|
|
|
|
- **Base Model**: unsloth/DeepSeek-R1-Distill-Llama-8B |
|
|
- **Training Method**: SFT (Supervised Fine-Tuning) |
|
|
- **Domain**: Medical reasoning and clinical knowledge |
|
|
- **Language**: English |
|
|
- **Quantization**: Q8_0 (gguf format for efficient inference) |
|
|
|
|
|
--- |
|
|
|
|
|
## 📚 Training Data |
|
|
|
|
|
The model was fine-tuned on: |
|
|
|
|
|
- **Dataset**: `FreedomIntelligence/medical-o1-reasoning-SFT` |
|
|
- **Language**: English |
|
|
- **Task**: Medical reasoning, clinical question-answering |
|
|
|
|
|
--- |
|
|
|
|
|
## 🚀 Usage Example |
|
|
|
|
|
> **Note:** The model is stored in `.gguf` format (quantized). You can load it using `unsloth` library. |
|
|
|
|
|
```python |
|
|
from unsloth import FastLanguageModel |
|
|
import torch |
|
|
|
|
|
# Load the quantized GGUF model |
|
|
model, tokenizer = FastLanguageModel.from_pretrained( |
|
|
"./unsloth.Q8_0.gguf", |
|
|
max_seq_length=2048, |
|
|
load_in_8bit=True, # optional depending on quantization |
|
|
) |
|
|
|
|
|
FastLanguageModel.for_inference(model) |
|
|
|
|
|
def generate(model, prompt, max_new_tokens=200): |
|
|
inputs = tokenizer(prompt, return_tensors="pt").to("cuda") |
|
|
|
|
|
with torch.no_grad(): |
|
|
outputs = model.generate( |
|
|
**inputs, |
|
|
max_new_tokens=max_new_tokens, |
|
|
do_sample=True, |
|
|
temperature=0.7, |
|
|
top_p=0.9, |
|
|
) |
|
|
|
|
|
return tokenizer.decode(outputs[0], skip_special_tokens=True) |
|
|
|
|
|
# Example prompt |
|
|
prompt = """### Instruction: |
|
|
"一个患有急性阑尾炎的病人已经发病5天，腹痛稍有减轻但仍然发热，在体检时发现右下腹有压痛的包块，此时应如何处理？" |
|
|
### Response: |
|
|
""" |
|
|
|
|
|
print(generate(model, prompt)) |
|
|
|
|
|
|
|
|
<think> |
|
|
这个病人已经有五天的急性阑尾炎了，腹痛虽然有所缓解，但他还是发烧，这让我有点担心。再看看他的体检结果，右下腹有压痛的包块，这就让我更有理由担心。嗯，既然有包块，而且病人还是发烧，应该是急性阑尾炎的急性症状。 |
|
|
|
|
|
想到这些，我觉得应该马上处理一下。最好的办法就是立即做腹部超声检查，看看包块的具体情况。超声检查可以帮助我们了解包块的位置、大小、形态以及周围组织的状态，这些信息对治疗非常重要。 |
|
|
|
|
|
哦，对了，包块如果是液体的，可能就不需要手术了。我们可以通过超声检查来确定这一点。这样就能避免不必要的手术创伤。 |
|
|
|
|
|
但是如果包块是固体的，特别是当它对周围组织有压迫作用时，手术就变得必要了。我们需要根据包块的具体情况来决定是否手术。 |
|
|
|
|
|
另外，包块的形态和位置也会影响手术的选择。我们需要根据这些因素来制定具体的手术方案。 |
|
|
|
|
|
嗯，总结一下，首先必须做一个腹部超声检查，确定包块的类型和具体情况。只有在确认包块是固体并且对周围组织有压迫作用的时候，才需要考虑手术。 |
|
|
|
|
|
好的，这个思路看起来很合理。现在就要把这些想法转化为具体的处理步骤。 |
|
|
</think> |
|
|
在这种情况下，建议首先进行腹部超声检查，以确定包块的类型和具体情况。根据检查结果： |
|
|
|
|
|
1. **如果包块是液体性包块**：通常不需要手术。可以通过药物治疗缓解症状，随后观察病情缓解情况。 |
|
|
|
|
|
2. **如果包块是固体性包块并且对周围组织有压迫作用**：这时手术就变得必要了。通常需要进行包块切除术，以解除压迫，缓解症状。 |
|
|
|
|
|
请尽快安排进行腹部超声检查，并根据检查结果制定最合适的手术方案。<｜end▁of▁sentence｜> |
