You can deploy this model and start earning money today!

abhishekrajdhar/
vidhi-v1

Name: vidhi-v1
Availability: InStock

Views

No views yet

Vidhi v1

Vidhi v1 is a lightweight healthcare-focused Small Language Model (SLM) trained on biomedical corpora and instruction-tuned for medical assistant behavior.

Features

Biomedical domain adaptation
Medical reasoning fine-tuning
Instruction-tuned assistant behavior
~64M parameters

Repository Structure

text

.
├── model.py
├── config.json
├── vidhi_medassist.pt
└── README.md

Usage

python

import torch
from model import Vidhi, VidhiConfig

device = "cuda"

config = VidhiConfig()

model = Vidhi(config)

state_dict = torch.load(
    "vidhi_medassist.pt",
    map_location=device
)

model.load_state_dict(state_dict)

model.to(device)
model.eval()

print("Vidhi loaded successfully")

Notes

Uses GPT-2 tokenizer (tiktoken)
Context length: 512
Custom PyTorch architecture
Research prototype only

Limitations

May hallucinate medical information
Not suitable for clinical deployment
No RLHF or retrieval augmentation yet

Author

Abhishek R. Dubey

abhishekrajdhar/vidhi-v1