Fix: Simplify invokeLLM to match the successful example exactly

Fix: Clean up code, remove fallback, and use direct Inference API for all models

Fix: Use direct Inference API for all models including DeepSeek for better stability

Fix: Use direct Inference API URL for abliterated model as per successful example

Fix: Change default model to official Qwen/Qwen2.5-72B-Instruct for stability

Fix: Explicitly use Featherless AI provider for abliterated models via HF Router and add fallback

Fix: Explicitly use Featherless AI provider for abliterated models via HF Router

Fix: Update LLM API endpoint from deprecated api-inference.huggingface.co to router.huggingface.co

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Temporary bypass: Disable login and session check for direct access

Fix: Provide fallback for missing VITE environment variables to prevent Invalid URL error

Fix: Add missing chart.js and react-chartjs-2 dependencies

Fix: Update Dockerfile and README metadata for proper deployment

Fix: Copy patches before install and update README metadata

Fix Dockerfile for Hugging Face build

Add Hugging Face metadata to README

Add Dockerfile for Hugging Face Spaces

Initial commit: HackingFactory v2 Enhanced with Self-Refining AI features