Google Git
Sign in
chromium/external/github.com/google-ai-edge/LiteRT-LM/HEAD
  1. 2bf48c9 Internal change by Google AI Edge · 11 hours ago upstream/main
  2. 92e1913 Make cache saving steps non-blocking in CMake CI workflow. by Google AI Edge · 13 hours ago
  3. 9714a66 refactor: consolidate Gemma 4 chat template test usages to models/gemma4 by Wai Hon Law · 16 hours ago
  4. 6eeede7 Apply fixes of https://github.com/google-ai-edge/litert-samples/pull/278 by Byungchul Kim · 17 hours ago
  5. ba82499 Remove absl::core_headers dependency from constrained decoding bitmap target. by Google AI Edge · 3 days ago
  6. 7ee7925 Add support for constrained decoding in the LiteRT MTP drafter. by Mohammadreza Heydary · 4 days ago
  7. 4a7f1c1 model(gemma4): add Gemma4 prompt templates and metadata configuration by Wai Hon Law · 4 days ago
  8. 56a9ac8 Bundle third-party license notices and generate SPDX SBOM for LiteRT-LM Swift artifacts. by Google AI Edge · 4 days ago
  9. a41b7c5 Finalize logit processor abstraction removal. by Mohammadreza Heydary · 4 days ago
  10. f1ec4db Simplify constraint handling in LLM executors by passing constrained_decoder directly. by Mohammadreza Heydary · 4 days ago
  11. a1d6367 Handle disabled profiling gracefully in LiteRT LM executor by Yi-Chun Kuo · 4 days ago
  12. 1349328 Internal changes only by Google AI Edge · 4 days ago
  13. 36d3fa5 Reverts a716c02275e8e09a449bffc19963334f7ee23162 by Google AI Edge · 4 days ago
  14. a8a8c44 Make constraint the root component and move toward removing logit processor abstraction. by Mohammadreza Heydary · 4 days ago
  15. d92aaa8 Expose Session and Conversation debug artifacts retrieval API in LiteRT-LM by Qidong Zhao · 4 days ago
  16. bca9d12 Update CMakeLists.txt to include Repetition Penalty Processor by Google AI Edge · 4 days ago
  17. 300faff refactor(model): remove unused namespace loop from Qwen3 chat template by Wai Hon Law · 4 days ago
  18. a716c02 LiteRT: return ScopedTensorBufferSpan from *TensorBufferAsSpan by Marissa Ikonomidis · 4 days ago
  19. 5fd90c0 Add loading by file descriptor to LiteRT LM Embedding Executor by Google AI Edge · 4 days ago
  20. 1b4d0ee Internal change by Google AI Edge · 5 days ago
  21. b12c62c model(qwen3): add Qwen3 prompt templates and metadata configuration by Wai Hon Law · 5 days ago
  22. 2978f98 Implement ComputeBitmap for CompositeConstraint with LogitMask fallback. by Mohammadreza Heydary · 5 days ago
  23. 931ae69 Fix SentencePiece header inclusion and OSS build on Windows. by Tenghui Zhu · 5 days ago
  24. af7951b Add support for the OpenAI /v1/embeddings HTTP API endpoint in LiteRT-LM CLI serve command using litert_lm.EmbeddingEngine. by Wai Hon Law · 5 days ago
  25. cd3ef22 Refactor sampler factory to use litert::Environment C++ type instead of raw C handle. by Chunlei Niu · 5 days ago
  26. 0252c01 Create auxiliary NPU components to allow code re-use by Matt Kreileder · 6 days ago
  27. 52011b8 Reverts 9fd961224613205e40c92f0b9584c9454d1fac46 by Tenghui Zhu · 6 days ago
  28. 06e235b Add model signature inspection utilities and selective compilation in LiteRT executors. by Matthew Chan · 6 days ago
  29. c08564c Use LogitsProcessor with CompositeConstraint in LiteRT LM runtime. by Mohammadreza Heydary · 6 days ago
  30. bbbe962 Refactor CMake build caching to separate dependencies from internal sources. by Google AI Edge · 6 days ago
  31. ac26f34 Add CompositeConstraint to combine multiple decoding constraints. by Mohammadreza Heydary · 6 days ago
  32. 00ba534 Implement logit processors as constraint. by Mohammadreza Heydary · 6 days ago
  33. 9fd9612 Internal change by Google AI Edge · 6 days ago
  34. 63772d8 Add audio preprocessor configuration to EmbeddingMetadata and update EmbeddingEngine to initialize AudioPreprocessorMiniAudio from metadata. by Matthew Chan · 6 days ago
  35. 96f487b Default normalize and insert_special_tokens to true in LiteRT-LM EmbeddingOptions. by Matthew Chan · 6 days ago
  36. ade63ee Introduce ComputeMask to the Constraint interface in constrained decoding. by Mohammadreza Heydary · 7 days ago
  37. 565ff04 fix: move capabilities C API to c/ directory and package in release zip by Wai Hon Law · 7 days ago
  38. ae1b008 Internal code update by Tenghui Zhu · 7 days ago
  39. 0ed32ec support fp16 logits in drafter. by Mohammadreza Heydary · 7 days ago
  40. 5ec54e6 test: minor fix to qwen3 chat template in test data by Wai Hon Law · 7 days ago
  41. c526842 Fix the include path and OSS build without special --config flags by Tenghui Zhu · 7 days ago
  42. 331e76c Add LogitMask interface and implementations for constrained decoding. by Mohammadreza Heydary · 7 days ago
  43. 23dd54c Move ccache configuration to the root CMakeLists.txt. by Google AI Edge · 7 days ago
  44. 710eff5 Made JSON schema serialization deterministic by enforcing sorted keys so prompts remain stable across runs, and enables engine-level constrained decoding via ResponseFormat (llguidance) so guided generation strictly conforms to the target schema. by Google AI Edge · 7 days ago
  45. 9589b6b Enable file backed file descriptor for NPU AOT by Google AI Edge · 8 days ago
  46. f9ccbd3 Fix use-of-uninitialized-value error in runtime debugger test. by Qidong Zhao · 8 days ago
  47. 1cdb9db fix: pass InputAudio directly to audio_preprocessor in EmbeddingEngineImpl by Wai Hon Law · 8 days ago
  48. d9f242b fix(kotlin): pass suppressTokens IntArray directly to JNI by Wai Hon Law · 10 days ago
  49. 8d1ce63 No public description by Wai Hon Law · 10 days ago
  50. cdeccd6 Overwrite the max_number_patches according to visual token budgets. by Yi-Chun Kuo · 10 days ago
  51. b4fbc38 Accept PCM audio data in Embedding layer by Google AI Edge · 11 days ago
  52. 94f85b8 Enforce matching version/date for API wheel in CLI build. by Google AI Edge · 11 days ago
  53. bd9754e Support optional synchronous execution in LiteRT-LM NPU runtime. by Salil Tambe · 11 days ago
  54. 2d5cfbd Changed TokenIdsToText, TokenIdsToTexts, and MergeTokenIds to accept by Google AI Edge · 11 days ago
  55. 30391d9 Implement RuntimeDebugger and Safetensors serialization utilities by Qidong Zhao · 11 days ago
  56. 1642cc3 Exports more WebGPU samplers by Google AI Edge · 11 days ago
  57. d8411e5 Update GitHub Actions workflow for Linux CMake builds and GCS caching. by Google AI Edge · 11 days ago
  58. 7c977ba Add mirror URLs for zlib dependency in WORKSPACE by Google AI Edge · 11 days ago
  59. 5c49038 feat: add embedding engine API for Swift by Wai Hon Law · 12 days ago
  60. abedc5e Consolidate key and value cache buffers into unified state buffers in LitertState. by Mohammadreza Heydary · 12 days ago
  61. 44d35f2 Add pyrefly suppressions by Hana Joo · 12 days ago
  62. a8a1539 feat: add embedding engine API for C and Python by Wai Hon Law · 12 days ago
  63. a719d96 Set cpu and gpu options in embedding LiteRT model executor by Yi-Chun Kuo · 12 days ago
  64. f11bba7 Support input overflow strategies in ComputeEmbedding. by Matthew Chan · 12 days ago
  65. 6439160 Fall back to Hugging Face for model downloads in ML CI workflows. by Google AI Edge · 13 days ago
  66. 7759d25 Internal change by Wai Hon Law · 13 days ago
  67. 1592f29 Updated LiteRT-LM README for v0.16.0 release. by Google AI Edge · 13 days ago
  68. a6f7eff Expose GetEmbeddingMetadata on EmbeddingEngine and EmbeddingEngineImpl. by Matthew Chan · 13 days ago
  69. 3bef5e5 Fix image placeholder matching in Lfm2DataProcessor and KV cache name matching by Salil Tambe · 13 days ago
  70. b9cbed9 Add BOS and EOS tokens to EmbeddingMetadata and EmbeddingEngineImpl. by Matthew Chan · 13 days ago
  71. e3feda9 NPU backend: Clear buffers post warmup run and init auxiliary buffers during create by Matt Kreileder · 13 days ago
  72. 4d27740 Initialize ImagePreprocessor in EmbeddingEngine using metadata vision parameters. by Matthew Chan · 13 days ago
  73. e0629e7 chore: version bump to 0.17.0 by Wai Hon Law · 13 days ago
  74. adc324e Use proto-defined AttentionMaskPolicy and AttentionMaskSettings directly. Fill ringbuffer input attention mask. by Yi-Chun Kuo · 13 days ago
  75. efc0f92 Add GetModelTypeName() function. by Google AI Edge · 13 days ago
  76. 636f41a Add insert_special_tokens option to EmbeddingOptions in EmbeddingEngine. by Matthew Chan · 14 days ago
  77. a20f320 Update LiteRT-LM Swift package to v0.16.0. by Google AI Edge · 14 days ago
  78. 29241cf Refactor streaming detokenization using BufferedStreamingDetokenizer. by Mohammadreza Heydary · 14 days ago
  79. 36238f3 Internal changes only. by Google AI Edge · 2 weeks ago
  80. 26bbb48 Fall back to the file descriptor for the GPU model cache key when the model by Google AI Edge · 2 weeks ago
  81. 672cd6c docs: document that max_output_tokens includes thinking tokens by Wai Hon Law · 2 weeks ago
  82. 5636ec5 Add patches merging option to image preprocessor and gemma4 config by Yi-Chun Kuo · 2 weeks ago
  83. d30be66 Reorder prompt tokens before image tokens in embedding_litert_lm_main. by Matthew Chan · 2 weeks ago
  84. 9bc7659 Add EmbeddingEngine C++ JNI implementation. by Google AI Edge · 2 weeks ago
  85. cba83bb Update LiteRT-LM Swift BUILD to export conversation.h by Google AI Edge · 2 weeks ago
  86. 3252476 Prepare LiteRT LM for ScopedTensorBufferSpan lock lifetimes by Marissa Ikonomidis · 2 weeks ago
  87. 5c72015 Internal Internal Internal by Google AI Edge · 2 weeks ago
  88. e533a5a Add Reset, Clone, Text Scoring, and BenchmarkInfo support to CachedSession. by Matthew Chan · 2 weeks ago
  89. a414022 Updated litertlm drafter and embedder to consume external weight by Google AI Edge · 2 weeks ago
  90. 55c2bc3 Updated litertlm vision encoder to consume external weight by Google AI Edge · 2 weeks ago
  91. 172ab84 Allow creating EmbeddingLookup with a litert::CompiledModel by Google AI Edge · 2 weeks ago
  92. a919ead Add qwen3-tts model type and inference by Yi-Chun Kuo · 2 weeks ago
  93. a3d29dc Enable YNNPACK optimization flag for Linux ARM64 builds in bazelrc. by Google AI Edge · 3 weeks ago
  94. 96b4819 Update dependencies of litert_lm by Byungchul Kim · 3 weeks ago
  95. fa814c1 Introduce two-phase CMake build orchestration for LiteRT-LM. by Google AI Edge · 3 weeks ago
  96. 0407a45 Refactor TtsEngine as the resource owner and TtsSession is a lightweight interface sharing compiled models with TtsEngine. by Yi-Chun Kuo · 3 weeks ago
  97. 800d1e6 Add utils to create compiled model and load files by Yi-Chun Kuo · 3 weeks ago
  98. fcf9daa Build with Ynnpack for ARM Linux Python Wheel by Terry Heo · 3 weeks ago
  99. c95b30a Rename unadapted/adapted audio embeddings in LiteRT LM. by Google AI Edge · 3 weeks ago
  100. fa187c4 Internal changes and clean up for chat template. by Google AI Edge · 3 weeks ago