Conversation
….1.4 Align SGLang's deps onto the apache-tvm-ffi 0.1.11 stack: - apache-tvm-ffi 0.1.9 -> 0.1.11 (tilelang 0.1.11 needs >=0.1.10; pin 0.1.11, since 0.1.12 breaks import, tile-ai/tilelang#2367) - tilelang 0.1.8 -> 0.1.11 - sgl-deep-gemm 0.1.3 -> 0.1.4 (python/pyproject.toml + docker/Dockerfile SGL_DEEP_GEMM_VERSION); sgl-deep-gemm 0.1.4 requires apache-tvm-ffi==0.1.11 This also fixes the current main inconsistency: sgl-deep-gemm (which requires apache-tvm-ffi==0.1.11) was pinned alongside apache-tvm-ffi==0.1.9. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
github-actions
Bot
added
the
dependencies
Collaborator
|
Hi @MartinHua , it will be included in this PR #29554 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Align SGLang's dependency pins onto the apache-tvm-ffi 0.1.11 stack:
apache-tvm-ffi0.1.9→0.1.11tilelang0.1.8→0.1.11sgl-deep-gemm0.1.3→0.1.4(python/pyproject.toml+docker/DockerfileSGL_DEEP_GEMM_VERSION)Why
maincurrently pinssgl-deep-gemm==0.1.3— whose metadata requiresapache-tvm-ffi==0.1.11— alongsideapache-tvm-ffi==0.1.9, which is an inconsistent (unresolvable) set. This PR fixes that and moves to the latestsgl-deep-gemm 0.1.4(alsorequires apache-tvm-ffi==0.1.11).tilelang 0.1.11requiresapache-tvm-ffi>=0.1.10; we pin0.1.11— not0.1.12, which aborts on import (tile-ai/tilelang#2367).Follows the already-merged build-time bump #28481 (
docker/sgl-deep-gemm.DockerfileTVM_FFI_VER → 0.1.11) and the DeepGEMM runtime pin bump (sgl-project/DeepGEMM#47).Verification
The 0.1.11 / fp4-MoE stack was validated end-to-end:
run_tests.sh+test_mega_moe_hopper(8×H100 SXM);--accuracy28/28 @ diff 0.0006.deepseek-ai/DeepSeek-V4-Flash(TP=4, fp4 MoE) → GSM8K 0.93–0.96, 0 invalid.🤖 Generated with Claude Code
CI States
Latest PR Test (Base): ❌ Run #28495104561
Latest PR Test (Extra): ❌ Run #28495104358