EkaCare Parrotlet-v-lite-4b

This is a purpose-built vision LLM specifically trained for the following downstream tasks on medical records in Indian healthcare context.

• Lab report parsing
• Digital prescription parsing
• Document classification
• PII extraction

A detailed description of this model can be obtained from this blog post.

Installation Requirements

To use this model, you need to install the following dependencies:

Python 3.10 and the following packages using pip:

pip install torch>=2.7.0 transformers>=4.52.0 pillow==11.3.0 huggingface_hub==0.36.0

Loading the model from Hugging Face Hub

from huggingface_hub import login
from transformers import AutoModel

login("hf_xxxxxxxxxxxxxxxxx")

repo_name = "ekacare/parrotlet-v-lite-4b"
model = AutoModel.from_pretrained(repo_name, trust_remote_code=True)

Parsing lab report

model.process("path/to/image", task="lab-report-parsing")

Task Types available

• lab-report-parsing
• prescription-parsing
• document-classification
• pii-parsing

License

This model is released under the MIT License, enabling broad use while maintaining attribution requirements.