Skip to content

nemotron_3_nano_nvfp4_fix_260201 #2620

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

Draft
malay-nagda wants to merge 5 commits into
base: main
Choose a base branch
from
Draft

nemotron_3_nano_nvfp4_fix_260201 #2620

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
0a0de66
nemotron_3_nano_nvfp4_fix_260201
malay-nagda Mar 3, 2026
5e3ee0f
b300 etp=None
malay-nagda Mar 3, 2026
ba9fbe8
gbs=128,ep=4,etp=n/a
malay-nagda Mar 3, 2026
17bfadb
no grouped gemm
malay-nagda Mar 3, 2026
b88b043
nvfp4 9004 flags
malay-nagda Mar 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions scripts/performance/perf_plugins.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -307,6 +307,11 @@ def _set_model_specific_environment_variables(
if "NVTE_NORM_BWD_USE_CUDNN" in executor.env_vars:
executor.env_vars.pop("NVTE_NORM_BWD_USE_CUDNN")

if model_recipe_name in ["nemotron_3_nano"] and compute_dtype == "nvfp4":
executor.env_vars["NVTE_NVFP4_DISABLE_RHT"] = "1"
executor.env_vars["NVTE_NVFP4_DISABLE_STOCHASTIC_ROUNDING"] = "1"
executor.env_vars["NVTE_NVFP4_DISABLE_2D_QUANTIZATION"] = "1"

def _set_layernorm_sm_margin(
self,
task: Union["run.Partial", "run.Script"],
Expand Down