-
Notifications
You must be signed in to change notification settings - Fork 25
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
porting sq to autoround #199
base: main
Are you sure you want to change the base?
Conversation
n1ck-guo
commented
Jul 24, 2024
- example
- api
- remove duplication process/code, increase running speed
- evaluation on few models
- tune alpha / exec order ?
Signed-off-by: n1ck-guo <[email protected]>
auto_round/smooth_quant/__init__.py
Outdated
@@ -0,0 +1,16 @@ | |||
#!/usr/bin/env python | |||
# -*- coding: utf-8 -*- |
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.
move it to algorithm_ext
absorb_to_layer = self.remove_unsupported_layers(model, absorb_to_layer, no_absorb_layers) | ||
return absorb_to_layer, no_absorb_layers | ||
|
||
def remove_unsupported_layers(self, model, absorb_to_layer, no_absorb_layers): |
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.
refer to autoawq to support handcrafted fused patten for popular models and use these configs first. graph trace has some limitations
"alpha_max": 1.0, | ||
"alpha_step": 0.1, | ||
"shared_criterion": "mean", | ||
"n_samples": 32, ##512 for cuda, 128 for cpu? |
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.
tune this parameters for gpu
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Please wait to merge until version 0.3 is released. |
Signed-off-by: n1ck-guo <[email protected]>