Repository navigation

[Perf][Qwen3-TTS]Cached Incremental Decode - #5202

Merged

linyueqian merged 28 commits into

vllm-project:mainfrom

sphinxkkkbc:feat/qwen3tts-cache-decode

Aug 11, 2026

Merged

[Perf][Qwen3-TTS]Cached Incremental Decode#5202
linyueqian merged 28 commits into
vllm-project:mainfrom
sphinxkkkbc:feat/qwen3tts-cache-decode

Commits on Jul 19, 2026
- add incremental decode for qwen3tts-icl mode, only support single-request
  sphinxkkkbc
  committed
  b8f496f
Commits on Jul 29, 2026
- split legacy branch into two explict banrches, revert forward_exact deletion
  sphinxkkkbc
  committed
  90e3eb1
Commits on Jul 30, 2026
- Merge branch 'vllm-project:main' into feat/qwen3tts-cache-decode
  sphinxkkkbc
  authored
  8098550
- add segmented graph wrapper
  sphinxkkkbc
  committed
  6050e9d
Commits on Jul 31, 2026
- add incremental decode for single request with dynamic ref_frame length
  sphinxkkkbc
  committed
  59840fb
- support batched graph decode
  sphinxkkkbc
  committed
  3bebab1
- remove hardcoded configuration
  sphinxkkkbc
  committed
  c326116
- support x-vector-only graph
  sphinxkkkbc
  committed
  c8f2f04
- fix x-vector-only bugs, add first chunk graph
  sphinxkkkbc
  committed
  93316e7
- stage_input_processor only emit new frames, remove capture_size configuration
  sphinxkkkbc
  committed
  d445236
Commits on Aug 1, 2026
Commits on Aug 2, 2026
Commits on Aug 4, 2026
- refactor graph state for maintenance
  sphinxkkkbc
  committed
  b8edbc8
- simplify comments
  sphinxkkkbc
  committed
  c1cd8f1
Commits on Aug 6, 2026
Commits on Aug 7, 2026
- pass initial_codec_chunk params internal
  sphinxkkkbc
  committed
  a0ccf58
Commits on Aug 9, 2026
- fix ci bugs: dummy_run cuda graph capture with dummy_weights crashes, need to be routed to forward_exact explictly, remove single request handling protocal, only keeps batch helpers, return list[to…
  sphinxkkkbc
  committed
  ffe679e
- remove unused codes for single request
  sphinxkkkbc
  committed
  08c3ba1
Commits on Aug 10, 2026
- Merge branch 'main' into feat/qwen3tts-cache-decode
  linyueqian
  authored
  e43c98a
Commits on Aug 11, 2026
- using internal_req_id for decoder_state cache, guard for left_context_size index misalignment, add regression for internal/external mismatch and index used for trimming
  sphinxkkkbc
  committed
  dc5a4e4
- Merge branch 'feat/qwen3tts-cache-decode' of github.com:sphinxkkkbc/vllm-omni into feat/qwen3tts-cache-decode
  sphinxkkkbc
  committed
  c63c2ef

Commits

Commits on Jul 19, 2026

Commits on Jul 29, 2026

Commits on Jul 30, 2026

Commits on Jul 31, 2026

Commits on Aug 1, 2026

Commits on Aug 2, 2026

Commits on Aug 4, 2026

Commits on Aug 6, 2026

Commits on Aug 7, 2026

Commits on Aug 9, 2026

Commits on Aug 10, 2026

Commits on Aug 11, 2026