Google Git
Sign in
chromium/external/github.com/google-ai-edge/LiteRT-LM/HEAD
  1. d9f192b Re-Implement sliding window attention mask when backed by a ringbuffer by Matt Kreileder · 22 hours ago upstream/main
  2. fe9a7bd internal changes ... by Yishuang Pang · 3 days ago
  3. df73046 Enable --ringbuffers-local-attention when possible (for models that support it) in the `litert-lm benchmark` CLI command by Google AI Edge · 3 days ago
  4. 529cf97 AsrEngine to run AsrSessions by Byungchul Kim · 3 days ago
  5. 909e1f1 Add BenchmarkInfo to EmbeddingEngine. by Matthew Chan · 3 days ago
  6. 465b314 Add --ringbuffers-local-attention flag to litert-lm CLI for models which support it by Google AI Edge · 3 days ago
  7. f6c83ef Add pre-run and post-run callbacks for model graph tensor buffers by Qidong Zhao · 3 days ago
  8. 5924eea Add memory-mapped file option to embedding_litert_lm_main.cc. by Matthew Chan · 3 days ago
  9. 4bd9e2a Fix missing LiteRT-LM benchmark profiling info by Terry Heo · 3 days ago
  10. 8ef6b80 Implement TdtDecoder by Byungchul Kim · 3 days ago
  11. 9cc919a Add MiniCpm5DataProcessor C++ data processor and metadata for MiniCPM5-1B. by Google AI Edge · 4 days ago
  12. 605e849 Update llm_litert_compiled_model_executor to enable proper sampler and host buffer creation when CPU/ GPU inference is going through the compiler plugin and dispatch API by Google AI Edge · 4 days ago upstream/release/v0.15
  13. 81f66c9 Internal change by Wai Hon Law · 4 days ago
  14. 849936d StatelessDecoder for Moonshine and Qwen3-ASR by Byungchul Kim · 4 days ago
  15. 09ea901 Support linear attention state buffers without sequence axis in LiteRT state. by Mohammadreza Heydary · 4 days ago
  16. e9bcf61 Reverts a88ee86db709420274b9508fe96fded80beb5cd2 by Matt Kreileder · 4 days ago
  17. 5982897 Internal change by Google AI Edge · 4 days ago
  18. f2195e9 Add MiniCPM5-1B model metadata and chat template to LiteRT-LM. by Google AI Edge · 4 days ago
  19. b6b91bb Enable state restoration for GPU-optimized single buffer cache. by Yi-Chun Kuo · 5 days ago
  20. a933349 Internal change by Wai Hon Law · 5 days ago
  21. d417931 Implement SpeechRecognizer by Byungchul Kim · 5 days ago
  22. 3ddad79 feat(python): migrate constrained decoding configuration to ConstrainedDecodingConfig dataclass by Wai Hon Law · 5 days ago
  23. a88ee86 Fix lock leaks and self-deadlock in LiteRT NPU runtime. by Salil Tambe · 5 days ago
  24. 6923204 Support input handling on GPU for OpenCL by Byungchul Kim · 5 days ago
  25. dce039d Enable a monotonic clock during benchmarking by Google AI Edge · 5 days ago
  26. 83734f4 Fix use-after-free heap corruption in NPU Environment::Create by Google AI Edge · 5 days ago
  27. 209c470 Rename namespace litert_lm to litert in omni. by Yi-Chun Kuo · 5 days ago
  28. f5ea5eb Internal change by Google AI Edge · 5 days ago
  29. 261c60d Introduce the notion of generic npu compiler plugin to allow graph compilation for CPU and GPU backend by Google AI Edge · 5 days ago
  30. 0f09100 Introduce TTS engine as the entry point for TTS inference. by Yi-Chun Kuo · 5 days ago
  31. 1a36e44 Ignore local state buffers when calculating maximum supported sequence size. by Mohammadreza Heydary · 5 days ago
  32. 6d6dfbe Implement Detokenizer with litert::support::Tokenizer by Byungchul Kim · 5 days ago
  33. 888dd37 Add file-based and stream-based text sources for LiteRT TTS. by Yi-Chun Kuo · 5 days ago
  34. 9b89a5e Add MelSpectroProcessor as AudioPreprocessor by Byungchul Kim · 6 days ago
  35. 210d379 Update dependencies of litert_lm by Mohammadreza Heydary · 6 days ago
  36. 5f96a06 Fix subtle sync issue in input handling on WebGPU by Byungchul Kim · 6 days ago
  37. 389d197 Add AudioSource from file. by Byungchul Kim · 6 days ago
  38. 3b3ffdf Declare param_tensor as an external tensor on Android by Byungchul Kim · 6 days ago
  39. 94e048f Update dependencies of litert_lm by Byungchul Kim · 7 days ago
  40. 9bb55f9 Fix Python 3.10 requirement and enable gfile pre-fetching in macOS Kokoro CI for litert_lm_cli. by Google AI Edge · 7 days ago
  41. a768079 Add EmbeddingEngine and embedding_litert_lm_main. by Matthew Chan · 7 days ago
  42. cff2f2c Support input handling on GPU for single KV cache case by Byungchul Kim · 7 days ago
  43. 7858613 Add missing embedding protos to litert_lm_builder shaded sources. by Matthew Chan · 7 days ago
  44. 101dee9 Add support for embedding metadata in the LiteRT-LM Python builder (`litertlm_builder.py`) and model inspection tool (`litertlm_peek.py`). by Matthew Chan · 7 days ago
  45. 61d2405 Introduce StateInterface for KV cache management. by Mohammadreza Heydary · 7 days ago
  46. 27f5801 Rename Linux presubmit check and add Android presubmit. by Google AI Edge · 7 days ago
  47. d3cc260 Publish LiteRT-LM API PyPI wheels using python3 -m uv. by Google AI Edge · 7 days ago
  48. b8d1ba7 Remove deprecated Windows and Linux CI build workflows. by Google AI Edge · 7 days ago
  49. fa95803 Don't set compiler plugin dir if NPU is not backend for any executors. by Byungchul Kim · 7 days ago
  50. 95f9e5d Adds API to retrieve audio embeddings by Google AI Edge · 7 days ago
  51. 7231f65 Introduce TTS session orchestration and pipeline stages for Omni TTS. by Yi-Chun Kuo · 10 days ago
  52. ee624ab Implement ASR session based on Stage and AsyncStageScheduler by Byungchul Kim · 10 days ago
  53. ccc945e Add Reset tests for static and dynamic LiteRT compiled model executors. by Mohammadreza Heydary · 10 days ago
  54. da6bdda SingleThreadedStageWithDeque template base class for session stage output queues by Byungchul Kim · 10 days ago
  55. df1af07 Constrain prefill work groups by remaining KV cache capacity. by Mohammadreza Heydary · 10 days ago
  56. bc1714f update test_lm file. by Mohammadreza Heydary · 10 days ago
  57. 69dce4e feat(cli): support activation_data_type in config schema and CLI commands by Wai Hon Law · 10 days ago
  58. cebb3ca Fix the misconfiguration in litert_util.cc. by Google AI Edge · 10 days ago
  59. 9604a2c Support `enable_ynnpack` and `num_cpu_threads` flag in litert_lm_main by Dillon Sharlet · 10 days ago
  60. 40cbf51 fix(python): annotate Content class attributes for Pyrefly by Wai Hon Law · 10 days ago
  61. 2dcaa83 fix(python): update the default value for cache_dir to None by Wai Hon Law · 10 days ago
  62. 171ae95 Drops usage of dynamic_runtime specific targets in litert_lm. by Chunlei Niu · 11 days ago
  63. 99811c2 Internal change and clean up by Google AI Edge · 11 days ago
  64. 11b5b93 Clean up LiteRTLM apple_fm backend by Google AI Edge · 11 days ago
  65. fa102fa Add EmbeddingExecutorBase interface and embedding_litert_compiled_model_executor. by Matthew Chan · 11 days ago
  66. 42d1035 Asynchronous API. by Byungchul Kim · 11 days ago
  67. f0a9a55 Test multi-turns with samplers which can handle inputs. by Byungchul Kim · 11 days ago
  68. c2ac06c No public description by Byungchul Kim · 12 days ago
  69. 091d175 Add multimodal support to Apple FM adapter merging from the external PR (https://github.com/john-rocky/swift-litert-lm/tree/main/pr/LiteRTLMFoundationModels). by Google AI Edge · 12 days ago
  70. e72a9a2 feat(cli): add $comment version metadata to config schema fields by Wai Hon Law · 12 days ago
  71. 8284640 Add support for different attention mask policies. by Yi-Chun Kuo · 12 days ago
  72. 4090766 feat(cli): make --ringbuffers-local-attention a hidden CLI flag by Wai Hon Law · 12 days ago
  73. caa0b0a Log executor mark durations in ms for LiteRT-LM. by Google AI Edge · 12 days ago
  74. 0416f58 docs: add version comments to public C API in engine.h by Wai Hon Law · 12 days ago
  75. d2d2a71 feat(cli): add support for --thinking and --thinking-budget in config and OpenAI handler by Wai Hon Law · 12 days ago
  76. 5df0c42 Introduce StateInterface for KV cache management. by Mohammadreza Heydary · 12 days ago
  77. 3f89e74 feat(cli): add --config option to use custom configuration file by Wai Hon Law · 12 days ago
  78. ad28d31 No public description by Byungchul Kim · 12 days ago
  79. 5f43d8a feat(cli): make --gpu-decode-steps-per-sync available in run command and config.json by Wai Hon Law · 12 days ago
  80. 83fa416 Import PR #2688: Support Data processor for vision encoder. by Salil Tambe · 13 days ago
  81. 61ff638 feat(cli): add --chat-template option to pack and unpack commands by Wai Hon Law · 13 days ago
  82. 533b08a Rename KV cache to state. by Mohammadreza Heydary · 13 days ago
  83. dc31f99 Add proto definitions for LLM executor metadata. by Mohammadreza Heydary · 13 days ago
  84. dc852ce Initialize litert_lm_cli Kokoro macOS pipeline by Google AI Edge · 13 days ago
  85. 8984994 No public description by Google AI Edge · 13 days ago
  86. feba9f7 Fix naming regression for weight cache by Marissa Ikonomidis · 13 days ago
  87. 0c43596 Expose GPU performance-sensitive flags to LiteRT-LM's PyThon layer and Benchmark CLI: by Google AI Edge · 13 days ago
  88. 8d33d4c Internal changes only. by Fengwu Yao · 13 days ago
  89. cad2499 feat(python): add use_ringbuffers_local_attention option in Python API and CLI by Wai Hon Law · 13 days ago
  90. ff43c68 Implement link-time dependency injection for LiteRT runtime. by Chunlei Niu · 13 days ago
  91. 9dcd6b6 Internal change by Google AI Edge · 13 days ago
  92. 54f1ae5 Use autosized ringbuffers in the LiteRT-LM js demo by Google AI Edge · 13 days ago
  93. df2e91f Add image and audio benchmark tests. by Google AI Edge · 13 days ago
  94. caaf3b1 Finalize LiteRT-LM PyPI Nightly Release Pipeline and Android Wheels by Google AI Edge · 14 days ago
  95. cc46545 Run pytest for dynamic linking build on Windows by Google AI Edge · 14 days ago
  96. c765ade Merged the external PR from (https://github.com/john-rocky/swift-litert-lm/tree/main/pr/LiteRTLMFoundationModels) about creating an adapter to allow Apple's foundation framework to use LiteRT-LM runtime. by Google AI Edge · 14 days ago
  97. ae89a59 Internal Change Only by Google AI Edge · 14 days ago
  98. f92cf36 Update dependencies of litert_lm by Tenghui Zhu · 14 days ago
  99. 54d80ae refactor(cli): drop jsonschema dependency in CLI config by Wai Hon Law · 14 days ago
  100. d051de3 Add option to error on invalid sampled token IDs in LiteRT-LM. by Yi-Chun Kuo · 14 days ago