Use official HuggingFace documentation approach to override model detection

Aggressive Docker enforcement - prevent HuggingFace Rust moshi-server override

Fix HuggingFace configuration - force Docker mode over Rust moshi-server

Major: Revert to full Moshi model with L4 GPU upgrade

Add comprehensive LMGen debugging and introspection

Add detailed debugging for lm_gen.step() returning None

Add C++ compiler for PyTorch Moshi compilation

Revert to working OMP_NUM_THREADS approach from SHA 9201e6d

Fix persistent OMP_NUM_THREADS override with unset + export

Fix LMGen API - use step() method instead of generate()

Add comprehensive transcription logging and implement actual LM generation

Fix NumPy array writable warning in audio processing

Fix OMP_NUM_THREADS being overridden by HuggingFace CPU_CORES

Switch to smaller Moshiko model for T4 GPU utilization

Fix device mismatch error - CUDA/CPU tensor mixing

Implement real audio processing in WebSocket instead of mock responses

Fix libgomp OMP_NUM_THREADS Invalid value error with explicit export

Fix CUDA memory issues and libgomp warning

Fix OMP_NUM_THREADS warning and restart loop issues

Fix OpenMP warning and application restart loop

Fix v1.3.6: Cache directory permissions - use /app/hf_cache instead of /.cache

Fix v1.3.3: Corrected Moshi import structure - use moshi.models directly

Fix v1.3.2: Runtime errors - Fixed Moshi imports, FastAPI lifespan, OpenMP config

Deploy v1.3.0: app.py - Official Moshi PyTorch STT implementation

Fix v1.1.2: app.py - Version bump for SDK configuration fix

Fix v1.1.1: app.py - transformers>=4.53.0 + 24kHz audio support

Fix Docker implementation: Upload app.py for real-time WebSocket STT streaming

Deploy v1.0.2: Working Gradio app with correct SHA d4fb4a2

Migrate working code: Deploy app.py v1.0.1 to correct Space

Upload app.py with huggingface_hub