Instructions to use jfan/litert-lm-web-runtime with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LiteRT
How to use jfan/litert-lm-web-runtime with LiteRT:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- LiteRT-LM
How to use jfan/litert-lm-web-runtime with LiteRT-LM:
# LiteRT-LM runs on various platforms (Android, iOS, Windows, Linux, macOS, IoT, Web/WASM) # and supports many APIs (C++, Python, Kotlin, Swift, JavaScript, Flutter). # For platform-specific integration guides, please refer to the official developer website: # https://ai.google.dev/edge/litert-lm # To try LiteRT-LM, the easiest way is to use our CLI tool. # 1. Install the LiteRT-LM CLI tool: pip install -U litert-lm # 2. Download and run this model locally: # See: https://ai.google.dev/edge/litert-lm/cli # A single .litertlm file in the repo is picked automatically; otherwise the CLI asks which one to run # (or pass its name right after the repo id). litert-lm run \ --from-huggingface-repo=jfan/litert-lm-web-runtime \ --prompt="Write me a poem"
- Notebooks
- Google Colab
- Kaggle
Commit History
Yesterday
-
Release @litert-lm /core 0.18.3 -c opt WebGPU WASM runtime (Piper HEAD CL 1014981939) with spatial_w_groups FC tiling and text scoring
Verified signature2ad809f
8 days ago
-
Release @litert-lm /core 0.18.3 WebGPU WASM runtime + Python 0.18.3 wheels (Lever 13c: Audio Adapter on WebGPU, + Lever 4b + 12B MQA)
Verified signaturec65c12c -
Release @litert-lm /core 0.18.3 WebGPU WASM runtime + Python 0.18.3 wheels (Lever 13c: Audio Adapter on WebGPU, + Lever 4b + 12B MQA)
Verified signatureb9bc9aa
9 days ago
-
Release @litert-lm /core 0.18.3 WebGPU WASM runtime + Python 0.18.3 wheels (lever4b multimodal prefill speedup, 12B MQA Metal shader fix, Metal & Linux wheels)
Verified signature0fe1694
16 days ago
-
Release @litert-lm /core 0.18.2 WebGPU WASM runtime + Python 0.18.2 wheels (subgroup-matrix, decide() API, Metal & Linux wheels)
Verified signature753bff8
19 days ago
-
Release @litert-lm /core 0.18.1 -c opt WebGPU WASM runtime with FusedAttention ViT self-attention (qnszmuyt) and untiled fusable attention chains (zvoytmmv): M5 Pro E4B vision encode -15% at 1120
Verified signature87f3f2b
22 days ago
-
Release @litert-lm /core 0.18.1 -c opt WebGPU WASM runtime with bounded-in-flight prefill command buffers (min(128, 16384/group_size)), createLegacyEngine/createStreaming compatibility, and prune wasm_dev_jank_v*
Verified signature09af975 -
Release @litert-lm /core 0.18.1 -c opt WebGPU WASM runtime with 1120 vision support and 57-CL stack optimizations
Verified signature10e35c7 -
Release @litert-lm /core 0.18.1 -c opt WebGPU WASM runtime with 1120 vision adapter selection fix (mszqroqq) and 10080 patches support
Verified signature1a8a435
23 days ago
-
Update LiteRT-LM Web runtime (0.18.1) with standardized 26B/31B vision support, trailing EOI rope mask fix, and SRQ FP16 capability guard
Verified signature77afd1b
24 days ago
-
Release @litert-lm /core 0.18.1 -c opt WebGPU WASM runtime with windowed [MTP_TELEMETRY]/[DECODE_TELEMETRY] (long-turn decay investigation, base CL 996998560) and prune main to one wasm set + one tgz
Verified signature300976b -
Release @litert-lm /core 0.18.1 -c opt WebGPU WASM runtime with L2 DP bucket calibration and L4 fused MoE
Verified signatureCalibrate prefill group_size DP cost model to eliminate multi-pass splits (L2) and fuse ExpertsRemapFrom + Transpose in MoE feed-forward layers (L4). Piper cl/989471319.6064f7b
25 days ago
-
Release @litert-lm /core 0.18.1 -c opt WebGPU WASM runtime with windowed [MTP_TELEMETRY]/[DECODE_TELEMETRY] (long-turn decay investigation, base CL 996998560) and prune main to one wasm set + one tgz
Verified signature7a3e421 -
Release @litert-lm /core@0.18.1 + -c opt WebGPU WASM runtime (Wave 2/4 stack: H2 rewind fallback + mutex fix, MTP token accounting, A0 AutoToolChat & vision defaults, A1 native profiler)
Verified signature4e679d1
26 days ago
-
Release final validated 66-CL stack -c opt WebGPU WASM runtime & @litert-lm /core@0.18.1 (M-ARCH, V12-fix, Track R, E2 non-blocking refill, P1 host parity shader drain)
Verified signature037940c -
Release @litert-lm /core@0.18.1 + -c opt WebGPU WASM runtime (all tracks R, V12-fix, M-ARCH, M5 V1..V6/E1..E2/P1..P3 merged)
Verified signature87c67e6
27 days ago
-
Release @litert-lm /core 0.18.1 WASM runtime & JS SDK (-c opt, Phase 3 T-D3a + T-D1 v2 + T-D3c + Task V)
Verified signature3f30428
28 days ago
-
Release @litert-lm /core 0.18.1 WASM runtime & JS SDK (-c opt, dc3acde07086)
Verified signaturebc3197e -
Release @litert-lm /core 0.18.1 WASM runtime & JS SDK (-c opt, e98354862e6c)
Verified signaturedb4c78a -
Release @litert-lm /core 0.18.1 WASM runtime & JS SDK (-c opt)
Verified signature6348755
29 days ago
-
Update LiteRT-LM Web runtime with prefill (P1-P7d) and decode (OPT-M2+M3 ScoreTokens staging & greedy drafter remap) optimizations
Verified signature83b1099 -
Update LiteRT-LM Web runtime with prefill (P1-P7d) and decode (OPT-M2+M3 ScoreTokens staging & greedy drafter remap) optimizations
Verified signature9fce354 -
Fix Gemma 4 12B MTP suppressed-token masking (scoring_post_processing_, centroid_constraint_mask, shared drafter config) and RejectAfterResample token_override preservation
Verified signature1b10bee -
Release @litert-lm /core 0.18.1: fix Apple Silicon prefill ringed_weights shader explosion, WebGPU device loss, and ReadBufferDataJs OOB
Verified signaturee05399f
30 days ago
-
Release -c opt WebGPU WASM runtime & @litert-lm /core@0.18.1: 64k CleanCacheOp/CopyCacheOp dispatch fix, decode subgraph deduplication, lazy vision/audio executor init, <empty_linear> warning fix, and live streaming benchmark cache
Verified signature36a37c6 -
Release @litert-lm /core@0.18.1 + -c opt WebGPU WASM runtime, fresh build from stack top cl/995335252 (includes cl/994475540, cl/995291160, cl/995335252)
Verified signaturea543cf3 -
Release -c opt WebGPU WASM runtime & @litert-lm /core@0.18.0-dev: Artisan decode FC linking, fake-quant dedupe, ring KV cache (+21% E2B standard decode on V100); stack synced to cl/995128901
Verified signature2239bdc -
Release @litert-lm /core 0.18.0 / 0.18.0-dev: RMSNorm gamma fusion (cl/994475540), cumulative vision compile/encode telemetry in litertlm_web.cc
Verified signatureadc4310
-
Release optimized -c opt WebGPU WASM runtime & @litert-lm /core@0.18.0-dev (clean ESM build, zero goog.*, EXP_WALLCLOCK_TUNE deadlock eliminated)
Verified signature166bee2 -
Release WebGPU WASM runtime & @litert-lm /core with baseline decode speedup (+40.77% on Gemma 4 E2B)
Verified signature3ea6403 -
Release latest WebGPU WASM runtime & @litert-lm /core (GEMV decode, bind group dispatch, graph fusions, and 26B real audio+MTP)
Verified signature5afe7c8 -
Release @litert-lm /core@0.18.0-dev (2026-10-06): drafter GPU config no longer inherits main-model MoE/laurel/altup config; built from workspace investigate_multimodal_web_support_jj @ cl/994039145
Verified signaturef0d711e
-
Release Gemma 4 MTP speculative acceptance gap closure runtime & @litert-lm /core 0.18.0
Verified signaturefa9d3f0 -
Fix MTP unseeded TOP_P greedy verification alignment + retain cached KV scales for InitShared
Verified signaturef9a10ee -
Fix 12B/26B/31B multimodal audio/vision initialization and enable realtime MTP acceptance telemetry during streaming
Verified signature8d04d9a -
Fix MTP shared KV scale cache retention across streaming LlmWebGpu::Init -> InitShared
Verified signature1795933 -
Update WASM runtime (-c opt) and @litert-lm /core 0.18.0-dev / 0.18.0 npm tarballs after stack slimming
Verified signature576ce9e
-
Publish clean ESM @litert-lm /core@0.18.0 bundle and -c opt WASM runtime
Verified signature3171277 -
Publish @litert-lm /core@0.18.1-dev ESM clean tsc build (fix goog is not defined)
Verified signaturec67ea5b -
Re-export @litert-lm /core@0.18.0-dev using standard tsc (no Google3 closure rules) & recompiled opt wasm
Verified signature2bf2459
-
Release @litert-lm /core@0.18.0-dev: MTP GPU PLE gather + full-vocabulary drafter head (1.44x MTP), audio FP32 fixes; synced to cl/992674170
Verified signaturef938177
-
Release @litert-lm /core@0.18.0-dev: fix PLE decode garbage, MTP quantized centroid scoring, deferred vision signature selection (vision_280 only)
Verified signaturec9ed261 -
Release @litert-lm /core@0.18.0-dev WASM runtime and tarball from latest 53-CL stack
Verified signature93ec636 -
Release @litert-lm /core@0.18.0-dev with cross-backend multimodal memory optimizations
Verified signaturec538263
-
Release WebGPU WASM runtime & @litert-lm /core v0.17.10-dev (unified cross-backend MTP & multimodal WebGPU runtime)
Verified signature7f48691
-
Release WebGPU WASM runtime & @litert-lm /core v0.17.10 (V100 WebGPU MTP & vision optimizations, live tool streaming, browser RGBA8 decode)
Verified signature1d5779d
-
Release LiteRT-LM Web Runtime 0.17.9-dev (clean 111-entry tarball + latest -c opt WASM)
Verified signaturee9432c9 -
Build latest WebGPU WASM runtime & @litert-lm /core (DraftTokens drafter_pos KV slot fix + Leviathan rejection sampling + top-p WGSL + media_cache_size_)
Verified signatureef70a47 -
Update WebGPU WASM runtime and core bundle with ConvGeneric fix for E4B
Verified signature39dae62 -
Update WebGPU WASM runtime and @litert-lm /core 0.17.9-dev with E2B/E4B vision & prefill fixes
Verified signaturea539ccc