Skip to content

Pull requests: vllm-project/tpu-inference

New pull request New
85 Open 1,207 Closed
85 Open 1,207 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

[Kernel][FusedMoe] Fix sync-barrier caused crash
#1342 opened Dec 19, 2025 by bythew3i Loading…
2
Precompile functions with large vocab_size tensors before allocating KV cache to avoid OOM ready ONLY add when PR is ready to merge/full CI is needed
#1341 opened Dec 19, 2025 by wenxindongwork Loading…
2
[DP] Reduce DP scheduling overhead via multiprocessing
#1340 opened Dec 19, 2025 by wenxindongwork • Draft
1
Add appache license. ready ONLY add when PR is ready to merge/full CI is needed
#1339 opened Dec 19, 2025 by QiliangCui Loading…
1
Use Topology Order to map KV cache P/D mapping ready ONLY add when PR is ready to merge/full CI is needed
#1338 opened Dec 19, 2025 by mrjunwan-lang Loading…
1
[CI] Add a vllm upstream integration pipeline
#1337 opened Dec 19, 2025 by weiyu0824 Loading…
10
Add GPQA Eval to Benchmarking
#1336 opened Dec 19, 2025 by AahilA Loading…
@patemotter
1
[Misc] Fix tpu platform init failure when vllm_config is not fully initialized
#1335 opened Dec 19, 2025 by sixiang-google Loading…
3
[Refactoring] use jax.shard_map instead of experimental one ready ONLY add when PR is ready to merge/full CI is needed
#1334 opened Dec 18, 2025 by lk-chen Loading…
2 of 3 tasks
1
[DeepSeek] Support TPU-Friendly Checkpoints + Add DeepSeek Testing
#1332 opened Dec 18, 2025 by jrplatin Loading…
1
[Kernel] Simplify the MLA bkv loading logic ready ONLY add when PR is ready to merge/full CI is needed
#1331 opened Dec 18, 2025 by yaochengji Loading…
WIP: Pd matching
#1329 opened Dec 17, 2025 by richardsliu • Draft
Apply temperature scaling before top-p
#1327 opened Dec 17, 2025 by oliverdutton Loading…
Attention DP for Torchax backend ready ONLY add when PR is ready to merge/full CI is needed
#1322 opened Dec 16, 2025 by wenxindongwork Loading…
10
add testcases to validate both prompt size less and greater than CPU RAM
#1317 opened Dec 15, 2025 by Sneha-at Loading…
21
[CI] This PR enhances testing of the CI procedures on both v6e and v7x. ready ONLY add when PR is ready to merge/full CI is needed
#1311 opened Dec 15, 2025 by dennisYehCienet Loading…
1
Refactor tuning for RPA HD64 kernel tuning to improve RPA kernel throughput ready ONLY add when PR is ready to merge/full CI is needed
#1308 opened Dec 14, 2025 by helloworld1 Loading…
3
Allow pytest to correctly discover all tests ready ONLY add when PR is ready to merge/full CI is needed
#1303 opened Dec 12, 2025 by wdhongtw Loading…
[do not merge ]Get all change files instead of last commit when bootstrap. ready ONLY add when PR is ready to merge/full CI is needed
#1299 opened Dec 12, 2025 by QiliangCui Loading…
1
[test do not review] ready ONLY add when PR is ready to merge/full CI is needed
#1298 opened Dec 12, 2025 by QiliangCui Loading…
1
[DRAFT] [DP][Bugfix] Fix bad sharding in non_dp case.
#1288 opened Dec 12, 2025 by py4 Loading…
3
[JAX][MoE] Integrate multiple MoE kernels in MoE modules
#1287 opened Dec 11, 2025 by bzgoogle Loading…
8
[multihost] Integrate expert parallelism to RayExecutor ready ONLY add when PR is ready to merge/full CI is needed
#1282 opened Dec 10, 2025 by Lumosis Loading…
2
[multihost] Use make_array_from_process_local_data to create global array instead of device_put
#1281 opened Dec 10, 2025 by Lumosis Loading…
4
[do not review][do not submit] ready ONLY add when PR is ready to merge/full CI is needed
#1277 opened Dec 10, 2025 by QiliangCui Loading…
1
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.