Skip to content
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

Change print() to logging in Quantizer so that the caller can have control #472

Merged
merged 2 commits into from
Jul 3, 2024
Merged

Change print() to logging in Quantizer so that the caller can have control #472

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
81d2417
Change print() to logging in Quantizer so that the caller can have co…
larryliu0820 Jul 3, 2024
f187931
Change Int8DynActInt4WeightQuantizer as well
larryliu0820 Jul 3, 2024
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
12 changes: 6 additions & 6 deletions torchao/quantization/GPTQ.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -621,7 +621,7 @@ def _create_quantized_state_dict(
out_features = mod.out_features
in_features = mod.in_features
# assert out_features % 8 == 0, "require out_features % 8 == 0"
print(f"linear: {fqn}, in={in_features}, out={out_features}")
logging.info(f"linear: {fqn}, in={in_features}, out={out_features}")

assert (
in_features % self.groupsize == 0
Expand All @@ -634,11 +634,11 @@ def _create_quantized_state_dict(
if self.padding_allowed:
from .utils import find_multiple
import torch.nn.functional as F
print(f"warning: {fqn} is padded to satisfy in_features % 1024 == 0")
logging.warn(f"warning: {fqn} is padded to satisfy in_features % 1024 == 0")
padded_in_features = find_multiple(in_features, 1024)
weight = F.pad(weight, pad=(0, padded_in_features - in_features))
else:
print(f"warning: {fqn} is skipped, int4 requires that in_features is 32, 64, or is divisible by 1024, " +
logging.warn(f"warning: {fqn} is skipped, int4 requires that in_features is 32, 64, or is divisible by 1024, " +
"and that groupsize and inner_k_tiles*16 evenly divide into it")
continue
(
Expand Down Expand Up @@ -960,7 +960,7 @@ def _create_quantized_state_dict(
out_features = mod.out_features
in_features = mod.in_features
# assert out_features % 8 == 0, "require out_features % 8 == 0"
print(f"linear: {fqn}, in={in_features}, out={out_features}")
logging.info(f"linear: {fqn}, in={in_features}, out={out_features}")

assert (
in_features % self.groupsize == 0
Expand All @@ -971,11 +971,11 @@ def _create_quantized_state_dict(
if self.padding_allowed:
from .utils import find_multiple
import torch.nn.functional as F
print(f"warning: {fqn} is padded to satisfy in_features % 1024 == 0")
logging.warn(f"warning: {fqn} is padded to satisfy in_features % 1024 == 0")
padded_in_features = find_multiple(in_features, 1024)
weight = F.pad(weight, pad=(0, padded_in_features - in_features))
else:
print(f"warning: {fqn} is skipped, int4 requires that in_features is 32, 64, or is divisible by 1024, " +
logging.warn(f"warning: {fqn} is skipped, int4 requires that in_features is 32, 64, or is divisible by 1024, " +
"and that groupsize and inner_k_tiles*16 evenly divide into it")
continue
(
Expand Down
Loading