Skip to content

Merge from main 2025-05-05 #18

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 50 commits into from
May 5, 2025
Merged

Merge from main 2025-05-05 #18

merged 50 commits into from
May 5, 2025

Conversation

Alexei-V-Ivanov-AMD
Copy link
Owner

@Alexei-V-Ivanov-AMD Alexei-V-Ivanov-AMD commented May 5, 2025 •
edited by github-actions bot
Loading

No description provided.

sstamenk and others added 30 commits May 1, 2025 11:51
@sstamenk
Remove duplicate code from dbrx.py (#17550)
04f2cfc
@CXIAAAAA
[Bug]change the position of cuda_graph_sizes in dataclasses (#17548)
173daac 
Signed-off-by: CXIAAAAA <[email protected]>
@Chenyaaang
[Misc][Tools][Benchmark] Publish script to auto tune server parameters (
9b70e2b 
#17207)

Signed-off-by: Chenyaaang <[email protected]>
@zixi-qi
[V1][Spec Decode] Apply torch.compile & cudagraph to EAGLE3 (#17504)
39c0813 
Signed-off-by: qizixi <[email protected]>
@mgoin
[Bugfix] Disable gptq_bitblas for <SM80 to fix GPTQ on V100/T4 (#17541)
24aebae 
Signed-off-by: mgoin <[email protected]>
@davidxia
[Doc] note that not all unit tests pass on CPU platforms (#17554)
afb12e4 
Signed-off-by: David Xia <[email protected]>
@LucasWilkinson
[Attention] MLA move o_proj q_proj into cuda-graph region (#17484)
afcb3f8 
Signed-off-by: Lucas Wilkinson <[email protected]>
@mgoin
[CI] Actually run tests/kv_transfer/test_disagg.py in CI (#17555)
292fc59 
Signed-off-by: mgoin <[email protected]>
@mgoin
Check if bitblas is installed during support check (#17572)
b4003d1 
Signed-off-by: mgoin <[email protected]>
@DarkLight1337
[Misc] Continue refactoring model tests (#17573)
f89d0e1 
Signed-off-by: DarkLight1337 <[email protected]>
@mgoin
Fix PixtralHF missing spatial_merge_size (#17571)
f192ca9 
Signed-off-by: mgoin <[email protected]>
@jerryzh168
Add pt_load_map_location to allow loading to cuda (#16869)
109e15a 
Signed-off-by: Jerry Zhang <[email protected]>
@Isotr0py
[Bugifx] Remove TritonPlaceholder from sys.modules (#17317)
9e2de9b 
Signed-off-by: Isotr0py <[email protected]>
@qthequartermasterman @Bryce1010
@Nan2018 @DarkLight1337
[Core] [Bugfix] Add Input Embeddings (#15428)
cc2a77d 
Signed-off-by: Andrew Sansom <[email protected]>
Signed-off-by: DarkLight1337 <[email protected]>
Co-authored-by: 临景 <[email protected]>
Co-authored-by: Bryce1010 <[email protected]>
Co-authored-by: Nan2018 <[email protected]>
Co-authored-by: Cyrus Leung <[email protected]>
Co-authored-by: DarkLight1337 <[email protected]>
@robertgshaw2-redhat
[BugFix] Fix Memory Leak (#17567)
c777df7 
Signed-off-by: [email protected] <[email protected]>
@DarkLight1337
[Misc] Rename assets for testing (#17575)
d754386 
Signed-off-by: DarkLight1337 <[email protected]>
@yangw-dev
add more pytorch related tests for torch nightly (#17422)
b8b0859 
Signed-off-by: Yang Wang <[email protected]>
@reidliu41
[doc] add the print result (#17584)
6d1479c 
Signed-off-by: reidliu41 <[email protected]>
Co-authored-by: reidliu41 <[email protected]>
@hmellor
Automatically tell users that dict args must be valid JSON in CLI (#1…
785d75a 
…7577)

Signed-off-by: Harry Mellor <[email protected]>
@DarkLight1337
[Security] Fix image hash collision (#17378)
99404f5 
Signed-off-by: DarkLight1337 <[email protected]>
@mgoin
Support W8A8 INT8 MoE for compressed-tensors (#16745)
868c546 
Signed-off-by: mgoin <[email protected]>
@reidliu41
[doc] miss result (#17589)
3a500cd 
Signed-off-by: reidliu41 <[email protected]>
Co-authored-by: reidliu41 <[email protected]>
@DarkLight1337
[Misc] Clean up input processing (#17582)
cb23495 
Signed-off-by: DarkLight1337 <[email protected]>
@hliuca
[Bugfix] fix tmp_out and exp_sums dimensions (#17438)
4c33d67 
Signed-off-by: Hui Liu <[email protected]>
@LucasWilkinson
[BugFix][Attention] Fix sliding window attention in V1 giving incorre…
0f87d8f 
…ct results (#17574)

Signed-off-by: Lucas Wilkinson <[email protected]>
@CalebDu
permute/unpermute kernel for moe optimization (#14568)
3e887d2 
Signed-off-by: Caleb_Du <[email protected]>
@Edwardf0t1
Add NVIDIA TensorRT Model Optimizer in vLLM documentation (#17561)
182f40e
@xw285cornell @houseroad
[Hardware][AMD] Improve OAM device ID + llama4 Maverick MOE tuning (#…
9352cdb 
…16263)

Signed-off-by: Lu Fang <[email protected]>
Co-authored-by: Lu Fang <[email protected]>
@zou3519
[easy] Print number of needed GPUs in skip message (#17594)
b90b085 
Signed-off-by: rzou <[email protected]>
@ehartford
fix typo in logging (#17605)
9b103a1
khluu and others added 19 commits May 2, 2025 18:54
@khluu
[release] Add command to clean up Docker containers/images in TPU rel…
3ec97e2 
…ease machine (#17606)
@liangfu
[Neuron][Build] Require setuptools >= 77.0.3 for PEP 639 (#17603)
22c6f63 
Signed-off-by: Liangfu Chen <[email protected]>
@22quinn
Update test requirements to CUDA 12.8 (#17576)
d47b605 
Signed-off-by: 22quinn <[email protected]>
@rasmith
[Quantizaton] [AMD] Add support for running DeepSeek int8 w8a8 MoE on…
e3d0a1d 
… ROCm (#17558)

Signed-off-by: Randall Smith <[email protected]>
@Chenyaaang
[Frontend][TPU] Add TPU default max-num-batched-tokens based on devic…
87baebe 
…e name (#17508)

Signed-off-by: Chenyaaang <[email protected]>
@tlrmchlsmth
[Build/CI] Upgrade CUTLASS to 3.9.1 (#17602)
c8386fa 
Signed-off-by: Tyler Michael Smith <[email protected]>
@gshtras
[Bugfix][ROCm] Using device_type because on ROCm the API is still tor…
a928424 
…ch.cuda (#17601)

Signed-off-by: Gregory Shtrasberg <[email protected]>
@DarkLight1337
[Core] Gate prompt_embeds behind a feature flag (#17607)
887d7af 
Signed-off-by: DarkLight1337 <[email protected]>
@Isotr0py
[Bugfix] Fix broken Qwen2.5-omni tests (#17613)
f66f1e0 
Signed-off-by: Isotr0py <[email protected]>
@DarkLight1337
[Misc] V0 fallback for --enable-prompt-embeds (#17615)
46fae69 
Signed-off-by: DarkLight1337 <[email protected]>
@hmellor
Add full API docs and improve the UX of navigating them (#17485)
d6484ef 
Signed-off-by: Harry Mellor <[email protected]>
@DarkLight1337
[Bugfix] Prioritize dtype in root config before checking text config (#…
2858830 
…17629)

Signed-off-by: DarkLight1337 <[email protected]>
@tlrmchlsmth
[Bugfix][Easy] Fix whitespace in shm_broadcast.py logging (#17635)
68e1ee0 
Signed-off-by: Tyler Michael Smith <[email protected]>
@chaunceyjiang
[Bugfix] fix KeyError on top logprobs are special tokens (#17637)
5394ad7 
Signed-off-by: chaunceyjiang <[email protected]>
@tlrmchlsmth
[Build/CI] Upgrade CUTLASS to 3.9.2 (#17641)
f62cad6 
Signed-off-by: Tyler Michael Smith <[email protected]>
@jinzhen-lin
[Kernel] some optimizations for dense marlin and moe marlin (#16850)
1d0c9d6 
Signed-off-by: Jinzhen Lin <[email protected]>
@Isotr0py
[Doc] Fix broken cuda installation doc rendering (#17654)
cc05b90 
Signed-off-by: Isotr0py <[email protected]>
@thomasjpfan
Use git-path commit in hook (#17616)
aea302b 
Signed-off-by: Thomas J. Fan <[email protected]>
@russellb
[Benchmarks] Remove invalid option under V1 engine (#17651)
d3efde8 
Signed-off-by: Russell Bryant <[email protected]>
@github-actions GitHub Actions
Copy link

github-actions bot commented May 5, 2025

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

@Alexei-V-Ivanov-AMD Alexei-V-Ivanov-AMD changed the title Re-base 2025-05-05 Merge from main 2025-05-05 May 5, 2025
@Alexei-V-Ivanov-AMD Alexei-V-Ivanov-AMD merged commit 0e6427e into Alexei-V-Ivanov-AMD:amd_experimental_all_tests_updated May 5, 2025
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.