ci: re-enable fp8 nightly benchmark configs (#22910)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
ishandhanani
2026-04-15 15:57:49 -07:00
committed by GitHub
co-authored by Claude Opus 4.6
parent e8c6e5466c
commit 761259448d
+19 -20
View File
@@ -9,26 +9,25 @@
# nightly-test-<runner>.yml workflow.
# Never edit workflow YAML files directly for these changes.
# TODO: re-enable after testing log analyzer (see follow-up PR)
# dsr1-fp8-gb200-dynamo-sglang:
# model: deepseek-ai/DeepSeek-R1-0528
# model-prefix: dsr1
# runner: gb200
# precision: fp8
# framework: dynamo-sglang
# multinode: true
# disagg: true
# seq-len-configs:
# - isl: 1024
# osl: 1024
# search-space:
# - conc-list: [1024, 2048, 4096, 6144]
# # https://github.com/NVIDIA/srt-slurm/blob/sglang-nightly-regression/recipes/gb200-fp8/1k1k/max-tpt.yaml
# config_file: recipes/gb200-fp8/1k1k/max-tpt.yaml
#
# - conc-list: [4096]
# # https://github.com/NVIDIA/srt-slurm/blob/sglang-nightly-regression/recipes/gb200-fp8/1k1k/ultra-tpt.yaml
# config_file: recipes/gb200-fp8/1k1k/ultra-tpt.yaml
dsr1-fp8-gb200-dynamo-sglang:
model: deepseek-ai/DeepSeek-R1-0528
model-prefix: dsr1
runner: gb200
precision: fp8
framework: dynamo-sglang
multinode: true
disagg: true
seq-len-configs:
- isl: 1024
osl: 1024
search-space:
- conc-list: [1024, 2048, 4096, 6144]
# https://github.com/NVIDIA/srt-slurm/blob/sglang-nightly-regression/recipes/gb200-fp8/1k1k/max-tpt.yaml
config_file: recipes/gb200-fp8/1k1k/max-tpt.yaml
- conc-list: [4096]
# https://github.com/NVIDIA/srt-slurm/blob/sglang-nightly-regression/recipes/gb200-fp8/1k1k/ultra-tpt.yaml
config_file: recipes/gb200-fp8/1k1k/ultra-tpt.yaml
dsr1-fp4-gb200-dynamo-sglang:
model: nvidia/DeepSeek-R1-0528-NVFP4-v2