Issues

Android GPU: `Engine.initialize()` freezes the app's screen, even on a background thread

google-ai-edge/LiteRT-LM#4315 · opened

on Oct 30, 2026

1

[NPU][MediaTek MT6989] Second message in an existing Conversation returns an empty reply (0.18.1, no tokens on turn 2)

google-ai-edge/LiteRT-LM#4256 · opened

on Oct 28, 2026

1

Feature request: cache encoded image/video tokens so prompts on the same image don't re-run the vision encoder

google-ai-edge/LiteRT-LM#4211 · opened

on Oct 27, 2026

1

Rewind leaves `last_prefill_token_id` stale, so the first decode/scoring step after it feeds the wrong token

google-ai-edge/LiteRT-LM#4195 · opened

on Oct 26, 2026

2

Gemma 4 12B: could the per-channel int4 checkpoint behind gemma-4-12B-it.litertlm be published? Re-quantizing the Q4_0 QAT weights to per-channel int4 breaks audio

google-ai-edge/LiteRT-LM#4185 · opened

on Oct 26, 2026

1

NPU executor: LlmMetadata.max_num_tokens larger than the compiled KV cache becomes the token limit (no clamp to the cache length)

google-ai-edge/LiteRT-LM#4184 · opened

on Oct 26, 2026

1

Gemma 4 E2B NPU bundle: CPU audio takes ~47 s per request for a 1.9 s clip on Snapdragon 8 Gen 3 (about 1.4 s with the CPU build)

google-ai-edge/LiteRT-LM#4183 · opened

on Oct 26, 2026

1

Expose max_input_length in CLI embedding-server configuration

google-ai-edge/LiteRT-LM#4179 · opened

on Oct 25, 2026

1

iOS Metal, v0.18.0: Qwen3.5 GPU programs fail to load from the cache (`DATA_LOSS`), so every engine creation compiles them again (about a minute)

google-ai-edge/LiteRT-LM#4178 · opened

on Oct 25, 2026

2

Docs: delivering a .litertlm model to Android users with Play AI packs

google-ai-edge/LiteRT-LM#4177 · opened

on Oct 25, 2026

2

Title: UTF-8 / French accent corruption with Gemma 4 E2B on Raspberry Pi 5 using LiteRT-LM

google-ai-edge/LiteRT-LM#4175 · opened

on Oct 24, 2026

3

EmbeddingGemma 2 audio embeddings are nondeterministic on SME2 CPUs (XNNPACK pf32 SME2 path)

google-ai-edge/LiteRT-LM#4145 · opened

on Oct 23, 2026

1

GPU engine create fails on 0.19.0.dev20261017+ for gemma-4-E2B-it: "Some ops are not accelerated. Add kLiteRtHwAcceleratorCpu..."

google-ai-edge/LiteRT-LM#4125 · opened

on Oct 22, 2026

1

[GPU] Static KV cache allocates full max_num_tokens for all layers, ignoring sliding-window layers (Gemma 4 E4B: ~76 KB/token, ~5x over-allocation)

google-ai-edge/LiteRT-LM#4096 · opened

on Oct 21, 2026

Gemma 4 12B on Windows Intel GPU (D3D12): no text when the prompt is under 512 tokens (prefill_128-only path)

google-ai-edge/LiteRT-LM#4083 · opened

on Oct 20, 2026

1

[JS] loadLiteRtLm() fails in module workers: module scripts do not support importScripts()

google-ai-edge/LiteRT-LM#4080 · opened

on Oct 20, 2026

1

litert_lm_builder: let callers keep uuid and creation_timestamp so that a rebuilt .litertlm can be byte-identical

google-ai-edge/LiteRT-LM#4064 · opened

on Oct 19, 2026

[FEATURE REQUEST] C API for LIteRT-LM Omni

google-ai-edge/LiteRT-LM#4058 · opened

on Oct 19, 2026

1

Benchmark `Init time` is about twice the engine creation time: `Init Total` is added to the phases it already contains

google-ai-edge/LiteRT-LM#4057 · opened

on Oct 19, 2026

3

C API (v0.16.0 prebuilt, Android GPU): `litert_lm_session_run_prefill` after a decode or `rewind_to_checkpoint` takes ~13 s regardless of length; also please ship Android C API prebuilts for 0.18 and expose checkpoints on the Kotlin Conversation and Session

google-ai-edge/LiteRT-LM#4054 · opened

on Oct 19, 2026

2

SamplerParams.k is not validated on the GPU path: k = 0 yields token id -1 and a tokenizer range error

google-ai-edge/LiteRT-LM#4052 · opened

on Oct 18, 2026

2

macOS WebGPU accelerator: Kokoro-82M omni_runner --backend=gpu outputs full-scale noise with exit 0 because weight-cache entries are wrapped as host-mapped buffers at a size Dawn rejects (CopyBufferToBuffer, "not aligned to 4096")

google-ai-edge/LiteRT-LM#4046 · opened

on Oct 17, 2026

3

@litert-lm/core EmbeddingEngine on CPU fails: "Failed to get external weight for buffer id …"

google-ai-edge/LiteRT-LM#4045 · opened

on Oct 17, 2026

9

EmbeddingEngine: API to count expanded input tokens without running inference

google-ai-edge/LiteRT-LM#4044 · opened

on Oct 17, 2026

2

[v0.18.0] iOS GemmaModelConstraintProvider prebuilt requires iOS 26.4 despite iOS 15 package support

google-ai-edge/LiteRT-LM#4043 · opened

on Oct 17, 2026

4

Navigation Menu

Repository navigation

Issues

Android GPU: `Engine.initialize()` freezes the app's screen, even on a background thread

[NPU][MediaTek MT6989] Second message in an existing Conversation returns an empty reply (0.18.1, no tokens on turn 2)

Feature request: cache encoded image/video tokens so prompts on the same image don't re-run the vision encoder

Rewind leaves `last_prefill_token_id` stale, so the first decode/scoring step after it feeds the wrong token

Gemma 4 12B: could the per-channel int4 checkpoint behind gemma-4-12B-it.litertlm be published? Re-quantizing the Q4_0 QAT weights to per-channel int4 breaks audio

NPU executor: LlmMetadata.max_num_tokens larger than the compiled KV cache becomes the token limit (no clamp to the cache length)

Gemma 4 E2B NPU bundle: CPU audio takes ~47 s per request for a 1.9 s clip on Snapdragon 8 Gen 3 (about 1.4 s with the CPU build)

Expose max_input_length in CLI embedding-server configuration

iOS Metal, v0.18.0: Qwen3.5 GPU programs fail to load from the cache (`DATA_LOSS`), so every engine creation compiles them again (about a minute)

Docs: delivering a .litertlm model to Android users with Play AI packs

Title: UTF-8 / French accent corruption with Gemma 4 E2B on Raspberry Pi 5 using LiteRT-LM

EmbeddingGemma 2 audio embeddings are nondeterministic on SME2 CPUs (XNNPACK pf32 SME2 path)

GPU engine create fails on 0.19.0.dev20261017+ for gemma-4-E2B-it: "Some ops are not accelerated. Add kLiteRtHwAcceleratorCpu..."

[GPU] Static KV cache allocates full max_num_tokens for all layers, ignoring sliding-window layers (Gemma 4 E4B: ~76 KB/token, ~5x over-allocation)

Gemma 4 12B on Windows Intel GPU (D3D12): no text when the prompt is under 512 tokens (prefill_128-only path)

[JS] loadLiteRtLm() fails in module workers: module scripts do not support importScripts()

litert_lm_builder: let callers keep uuid and creation_timestamp so that a rebuilt .litertlm can be byte-identical

[FEATURE REQUEST] C API for LIteRT-LM Omni

Benchmark `Init time` is about twice the engine creation time: `Init Total` is added to the phases it already contains

C API (v0.16.0 prebuilt, Android GPU): `litert_lm_session_run_prefill` after a decode or `rewind_to_checkpoint` takes ~13 s regardless of length; also please ship Android C API prebuilts for 0.18 and expose checkpoints on the Kotlin Conversation and Session

SamplerParams.k is not validated on the GPU path: k = 0 yields token id -1 and a tokenizer range error

macOS WebGPU accelerator: Kokoro-82M omni_runner --backend=gpu outputs full-scale noise with exit 0 because weight-cache entries are wrapped as host-mapped buffers at a size Dawn rejects (CopyBufferToBuffer, "not aligned to 4096")

@litert-lm/core EmbeddingEngine on CPU fails: "Failed to get external weight for buffer id …"

EmbeddingEngine: API to count expanded input tokens without running inference

[v0.18.0] iOS GemmaModelConstraintProvider prebuilt requires iOS 26.4 despite iOS 15 package support