Fix model paths: Use Candle-specific repository (-candle suffix)

Fix STT model file paths - correct tokenizer filenames

MAJOR FIX: Switch from conversation model to dedicated STT model

BREAKTHROUGH: Fix PAD token issue with correct STT start token

Enable audio boosting for quiet speech (RMS < 0.12)

Fix startup script syntax error - missing quote

Rebuild for A10G GPU: change CUDA compute capability from 75 to 86

Remove cross-attention from v0_1_vision config for Q8 model

Use v0_1_vision() config for exact Q8 model dimension match

Fix startup script to check correct Q8 model path

Fix Q8 model shape mismatch - let GGUF auto-configure dimensions

Fix GGUF model loading - HeaderTooLarge error resolved

Switch to moshiko-candle-q8 model to resolve Docker build memory constraints

Fix transformer config field names (validated) - v1.6.4

Fix LM config compilation error - v1.6.3

Fix model dimensions + pre-download moshiko-candle-bf16 models - v1.6.2

Fix audio tokenizer file path for moshiko-candle-bf16 - v1.6.1

Implement exact official Moshi backend configuration - v1.6.0

Switch to kyutai/moshiko-candle-q8 quantized model - v1.5.0

Fix Rust borrowing error in bounds checking - v1.4.40

Add defensive bounds checking for Moshi lm_state.step() - v1.4.39

v1.4.38: CRITICAL FIX - Moshi dual-stream configuration for STT

v1.4.37: Fix Moshi 0.6.3 panic with exact version pin

v1.4.36: Fix Rust server detection to reuse existing healthy servers

v1.4.35: Fix HuggingFace Spaces multi-process port binding issue

v1.4.34: Fix Rust server initialization in HuggingFace Spaces

v1.4.33: Add debugging tools to Docker runtime

v1.4.32: Add critical debug messages to trace HF Spaces app.py execution

v1.4.31: Move Rust server initialization to module level for HF Spaces

v1.4.30: Remove duplicate lifespan - single PRE-MOUNT Rust server only

v1.4.29: Fix FastAPI lifespan issue - start Rust BEFORE Gradio mounting

v1.4.28: Force clean rebuild with Moshi 0.6.1 to fix cached 0.6.3

v1.4.27: Revert to Moshi 0.6.1 + restore lifespan pattern

v1.4.26: RESTORE working FastAPI lifespan pattern from v1.4.19

v1.4.25: Move Rust server initialization back to __main__ block

v1.4.24: Implement module-level Rust server initialization for HF Spaces

fix: Remove unused FastAPI lifespan to prevent duplicate Rust servers

fix: Start Rust server directly in __main__ since Gradio mounting breaks lifespan

fix: Remove conflicting startup patterns preventing server from staying running

CRITICAL FIX: Remove duplicate Rust server initialization v1.4.20

VALIDATED FIX: Use FastAPI lifespan pattern for Rust server init v1.4.19

DEFINITIVE FIX: Configure Mimi for exactly 8 codebooks to match STT v0_1_one_way() v1.4.18

Fix HuggingFace Spaces app.py execution - move Rust server init to module level v1.4.17

Debug app.py execution - add startup debug message

Fix restart loop and update to v1.4.16

Use proper STT-only configuration: v0_1_one_way()

Fix array bounds: limit codes to 8 elements instead of 16

Fix hardcoded version in Dockerfile - now shows v1.4.13 with MOSHI PANIC FIXES

Add .dockerignore to force Rust rebuild - version 1.4.13

Force Rust binary rebuild with v1.4.13 - includes Moshi panic fix