Skip to content

Pull requests: vllm-project/tpu-inference

New pull request New
163 Open 1,832 Closed
163 Open 1,832 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Add ragged_conv1d for ragged chunked kernel integration for gdn attention ready ONLY add when PR is ready to merge/full CI is needed
#2099 opened Mar 31, 2026 by helloworld1 Loading…
1
[Environment variable override] Override VLLM_USE_AOT_COMPILE to False by default ready ONLY add when PR is ready to merge/full CI is needed
#2097 opened Mar 31, 2026 by jrplatin Loading…
2
Fix fork PR permissions
#2094 opened Mar 31, 2026 by ylangtsou Loading…
Call _align_hybrid_block_size in TpuPlatform ready ONLY add when PR is ready to merge/full CI is needed
#2090 opened Mar 31, 2026 by ShobhitBehl Loading…
3
[Draft] Add Qwen3.5 to CI/CD
#2086 opened Mar 30, 2026 by jrplatin • Draft
1
Set prompt_token_ids_cpu=None to match upstream interface change ready ONLY add when PR is ready to merge/full CI is needed
#2085 opened Mar 30, 2026 by pv97 Loading…
1
Pipeline psum and sc kernel
#2083 opened Mar 30, 2026 by clee1994 Loading…
env fix
#2077 opened Mar 28, 2026 by clee1994 Loading…
2
Fix compressed tensors moe test. ready ONLY add when PR is ready to merge/full CI is needed
#2076 opened Mar 28, 2026 by dmmolitor Loading…
4
Reorganization of ToC
#2064 opened Mar 27, 2026 by mtsokol • Draft
test
#2061 opened Mar 27, 2026 by meiyeh123 • Draft
1
Do not submit. ready ONLY add when PR is ready to merge/full CI is needed
#2053 opened Mar 26, 2026 by QiliangCui Loading…
1
[RPA] Add bidirectional attention support for multimodal
#2046 opened Mar 26, 2026 by kwang3939 Loading…
9
transcedentals cost in cost estimate ready ONLY add when PR is ready to merge/full CI is needed
#2044 opened Mar 26, 2026 by coolkp Loading…
6
Wire AWQ dense layers to use GMM V2 kernel for W4A16 matmul
#2038 opened Mar 26, 2026 by rohan-reddy Loading…
5 tasks done
[FP8][MoE] Move FP8 MoE weight requantization from CPU to TPU with cache clearing
#2028 opened Mar 25, 2026 by rohan-reddy Loading…
13
Hybrid MoE, combining EP and TP
#2001 opened Mar 23, 2026 by zhangamy-crypto Loading…
Implement ragged gather TC Pallas kernel to shard computation across devices
#1994 opened Mar 22, 2026 by bigeye • Draft
Fix #957: Support video input for Qwen2.5-VL
#1992 opened Mar 21, 2026 by codeXsidd Loading…
[Parallelism Support Matrix Tests] Replace flaky EP relative comparison with hardcoded absolute baseline ready ONLY add when PR is ready to merge/full CI is needed
#1984 opened Mar 20, 2026 by syhuang22 Loading…
@syhuang22
[Fused MoE] Use jax.nn.sigmoid ready ONLY add when PR is ready to merge/full CI is needed
#1980 opened Mar 20, 2026 by catswe Loading…
1
round kv cache to allow expert sharding
#1979 opened Mar 20, 2026 by khatwanimohit Loading…
1
[WIP] RPA KV update across bq
#1978 opened Mar 19, 2026 by helloworld1 • Draft
reorg support matrices for UX documentation Improvements or additions to documentation ready ONLY add when PR is ready to merge/full CI is needed
#1975 opened Mar 19, 2026 by jcyang43 Loading…
3
Add support for Qwen3-VL model via Torchax path
#1974 opened Mar 19, 2026 by muskansh-google Loading…
26
ProTip! Filter pull requests by the default branch with base:main.