108 lines
4.1 KiB
Markdown
108 lines
4.1 KiB
Markdown
---
|
||
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｜> |