---
base_model:
- mistralai/Mistral-7B-Instruct-v0.2
pipeline_tag: text-generation
---
#Introduction
# SRED Analysis Model

This model is fine-tuned for analyzing SRED (Scientific Research and Experimental Development) content, specifically for Box 242 analysis.

## Usage

Example input format:
```json
{
  "inputs": {
    "messages": [
      {
        "role": "system",
        "content": "You are an expert SRED technical writer analyzing Box 242 content."
      },
      {
        "role": "user",
        "content": "Analyze the following technological uncertainties..."
      }
    ]
  },
  "parameters": {
    "temperature": 0.7,
    "max_length": 1000
  }
}

#Getting Started
##Activate Virtual Environment
python3 -m venv .venv
source .venv/bin/activate
pip install --upgrade pip

# Ensure virtual environment is activated for pip install upgrade, otherwise there may be conflicts with your global environment

#For M2 Silicon Apple Chip
##install in this exact order:
pip3 install --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/cpu
pip install transformers==4.34.0
pip install datasets==2.14.0
pip install accelerate==0.24.0
pip install bitsandbytes==0.41.1
pip install wandb==0.15.12
pip install sentencepiece==0.1.99

##Finally install:
pip install peft==0.4.0

#For non-MX chips
##install in this order:
pip install transformers
pip install datasets
pip install accelerate
pip install bitsandbytes
pip install peft
pip install wandb
pip install torch
pip install sentencepiece

#For all users
pip install scipy
pip install easygui
pip install numpy==1.24.3
pip install python-dotenv

#Training setup
To create and upload required files to Hugging Face, run: python setup_files.py
To test your setup: run python test_setup.py
You will need to connect to WandB via WanB CLI
Update the .env file with your Hugging Face token, Hugging Face model name, and WandB token
The file should automatically push to Hugging Face

#Pushing just the model (no training)
If you need to just push the model, you can use: python push_model.py