Skip to content

Add FP4_MLP_ONLY option in truss config #1986

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

Merged
merged 4 commits into from
Oct 14, 2025
Merged

Add FP4_MLP_ONLY option in truss config #1986

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ff97433
Add FP4_MLP_ONLY option in truss config
AaryamSharmaBaseten Oct 13, 2025
1a068d8
Bump up truss version
AaryamSharmaBaseten Oct 13, 2025
042b208
Lint
AaryamSharmaBaseten Oct 14, 2025
0bdd210
Add check for FP4_KV
AaryamSharmaBaseten Oct 14, 2025
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
2 changes: 1 addition & 1 deletion pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
[project]
name = "truss"
version = "0.11.12"
version = "0.11.13.rc1"
description = "A seamless bridge from model development to model delivery"
authors = [
{ name = "Pankaj Gupta", email = "no-reply@baseten.co" },
Expand Down
4 changes: 3 additions & 1 deletion truss/base/trt_llm_config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,6 +68,7 @@ class TrussTRTLLMQuantizationType(str, Enum):
FP8_KV = "fp8_kv"
FP4 = "fp4"
FP4_KV = "fp4_kv"
FP4_MLP_ONLY = "fp4_mlp_only"


class TrussTRTLLMPluginConfiguration(PydanticTrTBaseModel):
Expand Down Expand Up @@ -713,7 +714,8 @@ def trt_llm_common_validation(config: "TrussConfig"):
"accelerators or newer (CUDA_COMPUTE>=89)"
)
elif trt_llm_config.build.quantization_type in [
TrussTRTLLMQuantizationType.FP4
TrussTRTLLMQuantizationType.FP4,
TrussTRTLLMQuantizationType.FP4_MLP_ONLY
] and config.resources.accelerator.accelerator in [
truss_config.Accelerator.H100,
truss_config.Accelerator.L4,
Expand Down
Loading