Fix: Use available CUDA 12.4.0 base image instead of non-existent 12.4.1

Switch to Docker build: removed gradio sdk, added Dockerfile for pure Docker deployment

Remove Docker files to use pure Gradio build on Hugging Face Spaces

Add gradio dependency and fix launch config for HF Spaces

Fix: Remove manual server config for HF Spaces compatibility

Fix: Enable share=True for Hugging Face Spaces compatibility

Add Hugging Face Space configuration in README header

Update repo with latest changes (excluding binary parquet files)

Fix setup.py: pin exact versions to match requirements.txt and prevent dependency conflicts

Upload folder using huggingface_hub

Add missing train_blank_slate.py training script

Remove chmod for train_blank_slate.py as file doesn't exist in repo

Fix Dockerfile: install torch separately, copy files before chmod

Fix Dockerfile: Install PyTorch separately to resolve accelerate dependency issue

fix: Add HF Spaces YAML configuration header

feat: Deploy complete NTF Space with Gradio UI, Docker, and auto-detect workers

initial commit