Skip to content

[SW-219274] - Changing the quant method name in lm-head (#150) #2132

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 3 commits into from
Feb 28, 2025
Merged

[SW-219274] - Changing the quant method name in lm-head (#150) #2132

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d71c787
[SW-219274] - Changing the quant method name in lm-head (#150)
nirda7 Feb 18, 2025
6c5b6e9
Update helper_modules.py
changwangss Feb 28, 2025
3dbfc1c
Merge branch 'master' into wangchang/pick
changwangss Feb 28, 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
14 changes: 7 additions & 7 deletions neural_compressor/torch/algorithms/fp8_quant/_quant_common/helper_modules.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,21 +233,21 @@ class PatchedParallelLMHead(PatchedModuleBase):
def __init__(self, mod, parent, mod_extra_config, *args, **kwargs):
super().__init__(mod, parent, mod_extra_config, *args, **kwargs)
# ParallelLMHead inherits from VocabParallelEmbedding (nn.module) which has a member called
# "linear_method" of type UnquantizedEmbeddingMethod that inherits from QuantizeMethodBase
# "quant_method" of type UnquantizedEmbeddingMethod that inherits from QuantizeMethodBase
# (both are not nn.module) and implement an "apply" method by using torch.nn.functional.linear
# (Apply the weights in layer to the input tensor.)
# ParallelLMHead's forward method should not be called because LMHead's weights should be used
# in the sampler. (The forward itself throws RuntimeError exception)
# So in order to quantize that linear_method we patch only the "apply" method.
# So in order to quantize that quant_method we patch only the "apply" method.
init_linear(self, mod_extra_config)
self.orig_linear_apply = types.MethodType(mod.linear_method.apply.__func__, self)
self.orig_linear_quant_apply = self.orig_mod.quant_method.apply
if self.quantization_mode in [QuantMode.QUANTIZE, QuantMode.LOAD]:
if self.use_qdq or self.fake_quant:
self.linear_method.apply = self.apply_qdq
self.quant_method.apply = self.apply_qdq
else:
self.linear_method.apply = self.apply_quant
self.quant_method.apply = self.apply_quant
elif (self.quantization_mode == QuantMode.MEASURE) or (self.quantization_mode == QuantMode.SHAPE):
self.linear_method.apply = self.apply_measure
self.quant_method.apply = self.apply_measure

def apply_quant(self, layer, x, bias):
qinput = self.quant_input(x)
Expand All @@ -273,7 +273,7 @@ def apply_qdq(self, layer, x, bias):

def apply_measure(self, layer, x, bias):
measure_input((x,), observer=self._mod_extra_config.inputs)
output = self.orig_linear_apply(layer, x, bias)
output = self.orig_linear_quant_apply(layer, x, bias)
measure_output((output,), self._mod_extra_config.outputs)
return output

Expand Down