Skip to content

docs(qwen3_8): update NVFP4 checkpoint reference to nvidia/Qwen3.8-27B-NVFP4 #1389

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
zhenshanx-nv merged 1 commit into from
Sep 22, 2026
+3 −3
Merged

docs(qwen3_8): update NVFP4 checkpoint reference to nvidia/Qwen3.8-27B-NVFP4 #1389

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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 families/qwen3_8/checkpoint_mapper.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -166,7 +166,7 @@ def _apply_block_scales(values: np.ndarray, scale_inv: np.ndarray) -> np.ndarray



# ModelOpt MIXED_PRECISION checkpoints (RadixArk/Qwen3.8-27B-NVFP4) carry two
# ModelOpt MIXED_PRECISION checkpoints (nvidia/Qwen3.8-27B-NVFP4) carry two
# schemes side by side, described by quantization_config.config_groups:
#
# FP8 attention and DeltaNet projections: float8_e4m3 weights with a single
Expand Down
4 changes: 2 additions & 2 deletions families/qwen3_8/quantization.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@

"""Qwen3.8-owned NVFP4 + FP8 TensorRT Q/DQ graph context.

RadixArk/Qwen3.8-27B-NVFP4 is a ModelOpt MIXED_PRECISION export that carries
nvidia/Qwen3.8-27B-NVFP4 is a ModelOpt MIXED_PRECISION export that carries
two quantization schemes side by side:

NVFP4 MLP projections (gate/up/down) and lm_head: E2M1 values packed two
Expand Down Expand Up @@ -474,7 +474,7 @@ def calibrate_qwen3_8_nvfp4(
if not scales:
raise RuntimeError(
"Qwen3.8 quantization calibration found no quantized tensors in "
"the checkpoint; is this a RadixArk/Qwen3.8-27B-NVFP4-style "
"the checkpoint; is this an nvidia/Qwen3.8-27B-NVFP4-style "
"checkpoint?"
)

Expand Down
Loading