feat(vllm): vllm gb200 dsv4 recipes - #103
Closed
alec-flowers wants to merge 52 commits into
Closed
alec-flowers wants to merge 52 commits into
alec-flowers wants to merge 52 commits into
Conversation
alec-flowers
changed the title
alec-flowers
force-pushed
the
aflowers/vllm-gb200-v0.20.0
branch
from
April 28, 2026 03:39
a48fe36 to
4acf4ea
Compare
alec-flowers
changed the title
alec-flowers
force-pushed
the
aflowers/vllm-gb200-v0.20.0
branch
from
April 28, 2026 03:55
707e933 to
c6df50a
Compare
alec-flowers
force-pushed
the
aflowers/vllm-gb200-v0.20.0
branch
from
April 29, 2026 01:11
406e5b4 to
7beaa58
Compare
alec-flowers
force-pushed
the
aflowers/vllm-gb200-v0.20.0
branch
from
April 29, 2026 01:32
7beaa58 to
653a652
Compare
Bump base_kv_events_port 5550 → 20000 and base_nixl_port 6550 → 21000 to avoid collisions with other services that bind in the 5000-7000 range. The previous values landed inside ranges used by random ephemeral / unprivileged services on shared compute nodes, causing sporadic 'address already in use' failures when multiple jobs landed on the same node. (cherry picked from commit 76ff18f on aflowers/gb200-dsv4-recipes)
Collaborator
|
Closing: the |
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Draft PR for the vLLM GB200 v0.20.0 branch.
Summary:
lm-evalbenchmark runner for GSM8K-style evals against an OpenAI-compatible chat endpoint.gsm8krunner untouched; this new path usespython3 -m lm_eval --model local-chat-completionsand does not require SGLang or an InferenceX workspace mount.src/srtctl/benchmarks/scripts/lm-eval/.recipes/vllm/deepseek-v4-pro/GB200/8k1k/disagg-gb200-1p4d-dep8-tp8-c256-c512-offload.yamlto runlm-evalwith the bundled GSM8K task andVALIDATE_EVAL_SCORES=true.Validation:
Qwen/Qwen2.5-0.5B-InstructOpenAI-compatible chat endpoint and ran the bundled script withEVAL_LIMIT=2,EVAL_NUM_FEWSHOT=0,EVAL_CONC=1; it producedmeta_env.json,results_*.json, andsamples_*.jsonlsuccessfully.bash -n src/srtctl/benchmarks/scripts/lm-eval/bench.shpassed.python3 -m py_compile src/srtctl/benchmarks/lm_eval.py src/srtctl/cli/do_sweep.py src/srtctl/benchmarks/scripts/lm-eval/validate_scores.pypassed.tests/test_benchmarks.py::TestLMEvalRunner,TestRunPostEval, andTestScriptsExist.UV_DEFAULT_INDEX=https://pypi.org/simple make checkpassed (635 passed, 2 skipped, 6 deselected).tydiagnostic insrc/srtctl/core/validation.pyis still emitted under the existing|| trueMakefile behavior.