Closed
Conversation
- configure device ids and ranks for data parallel
Co-authored-by: rebel-jonghewk <jonghewk@rebellions.in>
* rm requirements and update get optimum version * optimum rbln version * simplify workflow * add openai * fix * refactor: rm comments * rm condition for check-code-quality * debug patch * debug patch * simplify * rm requirements.txt * rm setup.py and requiremets * build without setup.py * fix if * fix dependency * [skip core] [skip worker] fix installing vllm-rbln in pytest-arc * [skip core] [skip worker] update runner * [skip worker][skip core][skip openai] pypi mirror server * [skip worker][skip core]for test * [skip core][skip worker][skip openai] * [skip core][skip worker][skip openai]separage each rsd cases * [skip worker][skip openai][skip core]fix dispatch * [skip worker][skip openai][skip core]fix dispatch * [skip worker][skip core][skip openai] fix * update runner * name fix * name fix * move out build vllm-rbln * fix path
* fix: rm requirements * rm v prefix * fix --------- Co-authored-by: Sungho Shin <87514200+rebel-shshin@users.noreply.github.com>
* enable profiler for optimum * precommit --------- Co-authored-by: rebel-jonghewk <jonghewk@rebellions.in>
* Update fsw pr ci * Update rh ci * Reflect comments * Update git pat * Fix github repository uppercase issue * Revert token
Co-authored-by: rebel-jaebin <jaebin@rebellions.ai>
* other(test): add unit test for attention backend.
* replace triton with torch_triton Signed-off-by: Jinseok Lee <jindol21@rebellions.ai>
* fix dp for v1 - remove DP padding support in v1 worker - add validation for DP implementation constraints in v1 worker - apply token mask to custom MOE kernel router logits - update default environment variables: - VLLM_RBLN_DP_IMPL: "dummy_prefill" -> "padded_decode" - VLLM_RBLN_USE_MOE_TOKENS_MASK: False -> True - fix DP metadata handling in forward context - add is_prefills field to RBLNFlashAttentionMetadata * fix test_rbln_envs.py - VLLM_RBLN_DP_IMPL should be padded_decode by default * fix DPMetadata for tokens mask - remove is_prefills field and related logic from DP metadata - fix get_tokens_mask() for non-DP case --------- Co-authored-by: rebel-jonghewk <142865404+rebel-jonghewk@users.noreply.github.com>
…rom performance tracking (#279)
* added torch_triton mode in RBLN_KERNEL_MODE Signed-off-by: Jinseok Lee <jindol21@rebellions.ai>
* fixed along with triton kernels in rebel_compiler Signed-off-by: Jinseok Lee <jindol21@rebellions.ai>
Signed-off-by: Jinseok Lee <jindol21@rebellions.ai>
Signed-off-by: Jinseok Lee <jindol21@rebellions.ai>
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
11 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.
🚀 Summary of Changes
This PR is for Release v0.9.5.
📌 Related Issues / Tickets
✅ Type of Change
feature)model)core)bug-fix)perf)refactor)docs)other): please describe🧪 How to Test
.........📸 Screenshots / Logs (if applicable)
📋 Checklist
💬 Notes