metadata

license: mit
language:
  - en
library_name: transformers
pipeline_tag: text-generation
tags:
  - medical
  - clinical

Model Card for Model ID

Model Details

Model Description

This model is a fine-tuned version of TheBloke/samantha-falcon-7B-GPTQ for text generation tasks in the medical domain.

Developed by: Pradhaph
Model type: Fine-tuned samantha-falcon-7B-GPTQ based model
Language(s) (NLP): English
License: MIT

Model Sources

Repository: 👉Click here👈
Demo: Available soon

Uses

Direct Use

This model can be used for text generation tasks in the medical domain, such as generating medical reports, answering medical queries, etc.

Downstream Use

This model can be fine-tuned for specific medical text generation tasks or integrated into larger healthcare systems.

Out-of-Scope Use

This model may not perform well on tasks outside the medical domain.

Bias, Risks, and Limitations

This model will requires more than 7.00GB GPU vram and 12.00GB CPU ram

How to Get Started with the Model

# Install dependencies
!pip install transformers==4.31.0 sentence_transformers==2.2.2
from transformers import AutoModelForCausalLM, AutoTokenizer

# 1. Load the model
loaded_model_path = r"path_to_downloaded_model"
model = AutoModelForCausalLM.from_pretrained(loaded_model_path)

# 2. Initialize the tokenizer
tokenizer = AutoTokenizer.from_pretrained(loaded_model_path)

# 3. Prepare input
context = "The context you want to provide to the model."
question = "The question you want to ask the model."
input_text = f"{context}\nQuestion: {question}\n"

# 4. Tokenize input
inputs = tokenizer(input_text, return_tensors="pt")

# 5. Model inference
with torch.no_grad():
    outputs = model.generate(
        **inputs,
        max_length=512,  # Adjust max_length as per your need
        temperature=0.7,  # Adjust temperature for randomness in sampling
        top_p=0.9,  # Adjust top_p for nucleus sampling
        num_return_sequences=1  # Number of sequences to generate
    )

# 6. Decode and print the output
generated_texts = [tokenizer.decode(output, skip_special_tokens=True) for output in outputs]
print("Generated Texts:")
for text in generated_texts:
    print(text)