HUD header: keep ROUND on one line, compact phase badge (▲ PEAK ▲) + NEW GAME so the whole row fits in normal and peak

HUD: shrink Help button to '?'; nowrap button labels so Help + New Game stay on one line

Add How-to-Play button + tip; rename to 'Mumbai Local' (drop PANIC in brand board + splash title)

Update README.md

Add LinkedIn post link

Rename to Mumbai Local; add demo video + blog links

README: accurate fallback wording + balance story anonymized (Model A)

Cleanup: remove internal handoff docs, BUILD.md, and dev eval logs (keep schemas/ runtime dep)

Update README: game premise, the bigger idea, and the full tech stack (ZeroGPU/Nemotron/Gradio Server+PixiJS)

GPU tuning: @spaces.GPU duration 90->60, max_new_tokens 768->320, remove diagnostic logging

Fix ZeroGPU anonymous-quota fumble: set window.supports_zerogpu_headers so @gradio/client forwards the x-ip-token (PRO quota instead of anon 60s)

DIAGNOSTIC: log the swallowed transformers-backend exception (full traceback)

Fix dispatcher fumbles: raw_decode first JSON object (small model repeats/think-wraps it; greedy regex spanned multiple braces -> no-op)

Warm VoxCPM2 at startup (under ZeroGPU emulation) + load_denoiser=False — fixes lazy-import CUDA-init error

Shim legacy generation output classes so the mamba-ssm hub kernel imports on transformers 5.x

Pin transformers==5.10.2 (5.11/5.12 dropped GreedySearchDecoderOnlyOutput the mamba-ssm kernel imports)

Nemotron-3-Nano fast Mamba path: native NemotronH (transformers>=5.8) + hub kernels, drop mamba_ssm wheel (fixes NVML fork assert)

Load Nemotron on CPU, move to GPU inside @spaces.GPU (fixes NVML fork assert)

Disable Nemotron thinking (enable_thinking=False) so the dispatcher emits clean JSON

Part 2: @app .api game endpoints + @gradio/client frontend (ZeroGPU auth-forwarded GPU calls)

Pin transformers==4.53.0 (nemotron_h modeling needs it; latest removed GreedySearchDecoderOnlyOutput)

Use lean mamba-ssm 2.2.5 (avoids 2.3.x triton>=3.5 conflict with torch 2.8)

Part 1: mamba-ssm wheel + module-level Nemotron load (real dispatcher on GPU)

Part 1: mamba-ssm wheel + module-level Nemotron load (real dispatcher on GPU)

Register a module-level @spaces.GPU function (HF ZeroGPU requires one at startup)

Bundle sprites + audio into the Space (no runtime bucket download)

Bundle assets (drop snapshot_download — its async teardown corrupted the server fd)

Non-daemon keepalive thread so the process stays resident under reload

Test python 3.10.13 (her-trace ver) + lazy model — isolate the serving keep-alive

Test python 3.10.13 (her-trace ver) + lazy model — isolate the serving keep-alive

Force module-level model load (her-trace keep-alive) + plain blocking launch

Keep main thread alive after launch (SSR on) so the Node proxy survives

Plain blocking launch with SSR on (match her-trace exactly)

Block main thread after launch (reload makes gr.Server launch non-blocking)

Disable GRADIO_WATCH_DIRS hot-reload (it breaks gr.Server launch blocking)

Self-contained root app.py = gr.Server + module-level launch (her-trace clone)

Self-contained root app.py = gr.Server + module-level launch (her-trace clone)

Expose demo=app for gradio reload mode (it launches the demo global)

app_file=frontend/server/app.py with SPACE_ID-gated module launch, ssr_mode=False

Remove importlib root app.py (app_file now frontend/server/app.py)

Launch gr.Server at MODULE level (HF gradio reload mode skips __main__)

Switch to gr.Server (Gradio Server mode) — fixes the 7860 double-bind

Pin gradio 6.16.0 + python 3.12 (match proven local-in-30s; 6.18 mount auto-launches -> double bind)

Fix Space entry: gradio host (demo.launch) instead of uvicorn (was double-binding 7860)

Deploy Mumbai Local Panic — Nemotron(ZeroGPU) dispatcher + VoxCPM2 voice

initial commit