42 lines
1.1 KiB
Markdown
42 lines
1.1 KiB
Markdown
|
---
|
||
|
|
license: openrail
|
||
|
|
---
|
||
|
|
|
||
|
|
# PMC_LLaMA
|
||
|
|
|
||
|
|
To obtain the foundation model in medical field, we propose [MedLLaMA_13B](https://huggingface.co/chaoyi-wu/MedLLaMA_13B) and PMC_LLaMA_13B.
|
||
|
|
|
||
|
|
MedLLaMA_13B is initialized from LLaMA-13B and further pretrained with medical corpus. Despite the expert knowledge gained, it lacks instruction-following ability.
|
||
|
|
Hereby we construct a instruction-tuning dataset and evaluate the tuned model.
|
||
|
|
|
||
|
|
As shown in the table, PMC_LLaMA_13B achieves comparable results to ChatGPT on medical QA benchmarks.
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
|
||
|
|
## Usage
|
||
|
|
|
||
|
|
```python
|
||
|
|
import transformers
|
||
|
|
import torch
|
||
|
|
|
||
|
|
tokenizer = transformers.LlamaTokenizer.from_pretrained('axiong/PMC_LLaMA_13B')
|
||
|
|
model = transformers.LlamaForCausalLM.from_pretrained('axiong/PMC_LLaMA_13B')
|
||
|
|
|
||
|
|
sentence = 'Hello, doctor'
|
||
|
|
batch = tokenizer(
|
||
|
|
sentence,
|
||
|
|
return_tensors="pt",
|
||
|
|
add_special_tokens=False
|
||
|
|
)
|
||
|
|
with torch.no_grad():
|
||
|
|
generated = model.generate(
|
||
|
|
inputs = batch["input_ids"],
|
||
|
|
max_length=200,
|
||
|
|
do_sample=True,
|
||
|
|
top_k=50
|
||
|
|
)
|
||
|
|
print('model predict: ',tokenizer.decode(generated[0]))
|
||
|
|
```
|
||
|
|
|