-
Notifications
You must be signed in to change notification settings - Fork 177
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.
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
Add int4 weight-only embedding QAT #947
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/947
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit cbf980e with merge base 8af72fc (): This comment was automatically generated by Dr. CI and updates every 15 minutes. |
1cc1b42
to
3647c75
Compare
Based on changes in D62664322 by Tijmen Blankevoort.
3647c75
to
cbf980e
Compare
def forward(self, x): | ||
from torchao._executorch_ops import _quantized_decomposed_dequantize_per_channel_group_wrapper | ||
qmin, qmax = _get_qmin_qmax(self.bit_width) | ||
w_dq = _quantized_decomposed_dequantize_per_channel_group_wrapper( |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
is this lowered to executorch?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yeah
Based on changes in pytorch#886 by @TiRune
Based on changes in #886 by @TiRune