BioMistral-Clinical-7B

Model Card for Model ID

More code details can be found at Github: https://github.com/Incredible88/BioMistral-Clinical-7B

How to use

Loading the model from Hunggingface:

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ZiweiChen/BioMistral-Clinical-7B")
model = AutoModelForCausalLM.from_pretrained("ZiweiChen/BioMistral-Clinical-7B")

Lightweight model loading can be used - using 4-bit quantization!

!pip install -q -U bitsandbytes
!pip install -q -U git+https://github.com/huggingface/transformers.git
!pip install -q -U git+https://github.com/huggingface/peft.git
!pip install -q -U git+https://github.com/huggingface/accelerate.git

from transformers import  AutoTokenizer, BitsAndBytesConfig, AutoModelForCausalLM
import torch

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_use_double_quant=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16
)

tokenizer = AutoTokenizer.from_pretrained("ZiweiChen/BioMistral-Clinical-7B")
model = AutoModelForCausalLM.from_pretrained("ZiweiChen/BioMistral-Clinical-7B", quantization_config=bnb_config)

How to Generate text:

model_device = next(model.parameters()).device

prompt = """
### Question:

How to treat severe obesity?

### Answer:
"""
model_input = tokenizer(prompt, return_tensors="pt").to(model_device)

with torch.no_grad():
    output = model.generate(**model_input, max_new_tokens=100)
    answer = tokenizer.decode(output[0], skip_special_tokens=True)
    print(answer)

Incremental learning

The process of incremental learning:
The training process records:

Clinical Scenario Analysis

More informative answer than BioMistral-7B:

Supervised Fine-tuning Benchmark

CAUTION! Both direct and downstream users need to be informed about the risks, biases, and constraints inherent in the model. While the model can produce natural language text, our exploration of its capabilities and limitations is just beginning. In fields such as medicine, comprehending these limitations is crucial. Hence, we strongly advise against deploying this model for natural language generation in production or for professional tasks in the realm of health and medicine.