Skip to content

run forward pass even for empty subset to produce correct layer outputs #2161

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 2 commits into from
Nov 2, 2025
Merged

run forward pass even for empty subset to produce correct layer outputs #2161

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8156bd6
run forward pass even for empty subset to produce correct layer outputs
avtc Nov 1, 2025
465881f
add todo comment
avtc Nov 2, 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
33 changes: 22 additions & 11 deletions gptqmodel/looper/stage_subset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,17 +88,28 @@ def run_subset_stage(
layer_module=module,
)

if len(subset) == 0:
if logger.isEnabledFor(logging.DEBUG):
logger.debug(
"StageSubset: layer=%s subset=%s/%s processor=%s produced empty subset (names=%s)",
layer_index,
subset_index + 1,
subset_total,
processor_name,
subset_names,
)
return SubsetStageResult(processed_subset={}, layer_inputs=layer_inputs, forward_context=None)
# TODO FIXME: If a full layer has no module to quantize a simple forward() is enough and output is captured
# to be used as next layer's input. So one pass forward (entire layer simple forward wihout need of dealing
# with subset loops and micro forward loops, just full layer, usally XXXDecodeLayer.forward().
# So output = current_layer.forward() is enough or sometimes just calling the layer callable like layer()
# which same as layer.forward().
#
# Assume layer 2 has no modules to quantize. At beginniing loop for layer 2, we have layer_output
# from completed forward_replay() of layer 1. Then pass this to layer 2 (as a whole) as layer_input
# and store ouput, then immediately loop to layer 3 without any further subset work that is only necessary
# if we need to quantize part of a layer.
#
# if len(subset) == 0:
# if logger.isEnabledFor(logging.DEBUG):
# logger.debug(
# "StageSubset: layer=%s subset=%s/%s processor=%s produced empty subset (names=%s)",
# layer_index,
# subset_index + 1,
# subset_total,
# processor_name,
# subset_names,
# )
# return SubsetStageResult(processed_subset={}, layer_inputs=layer_inputs, forward_context=None)

if is_awq_processor:
logger.info(
Expand Down