Ben-Cars0n/pharma-tinyllama-domain-lora

library_name: peft license: mit language:

• —en tags:
• —lora
• —tinyllama
• —fine-tuning
• —medical ---

Pharma TinyLlama Domain LoRA

This repository contains a LoRA (Low-Rank Adaptation) fine-tuned TinyLlama model for the pharmaceutical domain.

Contents

• —adapter_model.safetensors: The LoRA adapter weights.
• —adapter_config.json: Configuration for the LoRA adapter.
• —tokenizer.json, tokenizer_config.json: The tokenizer used for training.
• —processed_data/pdf_pages_raw.jsonl: Raw extracted text from PDF pages.
• —processed_data/pharma_paragraph_process.jsonl: Cleaned and processed paragraph data.

Usage

To load and use this model with the adapter, you can follow the Hugging Face PEFT library documentation.

Resources

• —The merged model, which combines the base TinyLlama model with the LoRA adapter weights, is available in the merged_model directory within this repository. This can be used for direct deployment.