Update latest_model.pt with multi-head fixes

Update latest_model.pt with multi-head fixes

Update eduforge_runtime.py with multi-head fixes

Update latest_model.pt with multi-head fixes

Update train_brain.py with multi-head fixes

Update eduforge_runtime.py with multi-head fixes

Update app.py with multi-head fixes

Update latest_model.pt with multi-head fixes

Update train_brain.py with multi-head fixes

Update hybrid_bridge.py with multi-head fixes

Fix: harden tutoring response keyword matching

Fix: smart tutoring responses, proxy config, fine-tuned model integration

Fix Dockerfile paths for frontend build context

Finalized project with robust demo mode and fixed trainer configuration

Fix: Moved __future__ import to top of engine.py to resolve SyntaxError

Data Prep: Successfully converted training data to JSONL format for AutoTrain

Final Conversational Fix: Added One-Shot examples and hard-coded regex stripping for 'Regarding' prefixes

RL Brain Upgrade: Forced update of latest_model.pt after 27k sample training

RL Brain Upgrade: Trained DQN on 27k+ new student transitions

Conversational fix: Improved system prompt to handle greetings and avoid meta-narrative

Force-added expanded training samples dataset

Added missing groq dependency

Updated Node.js version to 20 for Next.js compatibility

Production readiness: Enabled static export and integrated frontend serving in app.py

Fixed Dockerfile syntax error

Initial clean deployment to Hugging Face