mimi__mimi_dec_tx_fp16 — LiteRT.js demo

Source	https://huggingface.co/litert-community/Mimi
License	cc-by-4.0
Model file	`f2f0421f2ac6d3fb17892450bf765ca57cfaea2b58b5af79955997d41c2f1db4.tflite` — downloaded from the source at runtime; this site hosts no weights

Run it in your browser

Measured sweep results

Every number is labeled with the environment that produced it — headless desktop Chromium is not a user’s phone, and results are only meaningful together with their environment.

Backend	Status	Full delegation	Output match	Latency p50 (ms)	Environment	Verified	Provenance
`wasm_xnnpack`	pass	—	—	40.158	mac-studio-m4-max · chromium 151.0.7922.34 headless · macOS · @litertjs/core 2.5.3	2026-08-11	measured
`webgpu_mldrift`	output_mismatch	yes	fail	4.683	mac-studio-m4-max · chromium 151.0.7922.34 headless · macOS · @litertjs/core 2.5.3	2026-08-11	measured

Build this yourself

One-line prompt for your AI coding assistant (uses the litert-cookbook web skill):

Use the litert-cookbook web skill to build a LiteRT.js browser demo for mimi__mimi_dec_tx_fp16: load the model from https://huggingface.co/litert-community/Mimi/resolve/main/f2f0421f2ac6d3fb17892450bf765ca57cfaea2b58b5af79955997d41c2f1db4.tflite (license: cc-by-4.0) and run it with @litertjs/core.

Reproduce this with an agent

One-line prompt for Claude Code / Gemini CLI — re-runs the same sweep verification for this model, locally:

In the edge-compat repo, verify mimi__mimi_dec_tx_fp16 with the sweep harness: write /tmp/catalog.csv with the header line model_id,source,input_spec,license,notes and the single row mimi__mimi_dec_tx_fp16,https://huggingface.co/litert-community/Mimi/resolve/main/f2f0421f2ac6d3fb17892450bf765ca57cfaea2b58b5af79955997d41c2f1db4.tflite,1x512x25:float32,cc-by-4.0, — then run: cd web/sweep && npm ci && npm run sweep -- --catalog /tmp/catalog.csv --out out/ — and compare out/mimi__mimi_dec_tx_fp16.json with the sweep table on this page.