[GH-ISSUE #253] CLIP image_encoder as backbone #222

New issue

Closed

opened 2026-10-31 21:12:37 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-31 21:12:37 +00:00

Owner

Originally created by @iasonasxrist on GitHub (Jun 15, 2022).
Original GitHub issue: https://github.com/openai/CLIP/issues/253

I have been trying to implement the model.py to separate the image encoder from text but i got plenty of errors kind of :

self.positional_embedding = nn.Parameter(torch.empty(self.context_length, transformer_width))

raise AttributeError("'{}' object has no attribute '{}'".format(
AttributeError: 'CLIP' object has no attribute 'context_length'

Any ideas how to override them?

Originally created by @iasonasxrist on GitHub (Jun 15, 2022). Original GitHub issue: https://github.com/openai/CLIP/issues/253 I have been trying to implement the model.py to separate the image encoder from text but i got plenty of errors kind of : self.positional_embedding = nn.Parameter(torch.empty(self.context_length, transformer_width)) raise AttributeError("'{}' object has no attribute '{}'".format( AttributeError: 'CLIP' object has no attribute 'context_length' Any ideas how to override them?

bsakaguchi closed this issue

2026-10-31 21:12:37 +00:00

bsakaguchi commented

2026-10-31 21:12:38 +00:00

Author

Owner

@lr19960813 commented on GitHub (Mar 27, 2023):

I have been trying to implement the model.py to separate the image encoder from text but i got plenty of errors kind of :

self.positional_embedding = nn.Parameter(torch.empty(self.context_length, transformer_width))

raise AttributeError("'{}' object has no attribute '{}'".format( AttributeError: 'CLIP' object has no attribute 'context_length'

Any ideas how to override them?

Hi, I encountered the same problem as you, how did you solve it, looking forward to your reply. Thanks a lot!

@lr19960813 commented on GitHub (Mar 27, 2023): > I have been trying to implement the model.py to separate the image encoder from text but i got plenty of errors kind of : > > self.positional_embedding = nn.Parameter(torch.empty(self.context_length, transformer_width)) > > raise AttributeError("'{}' object has no attribute '{}'".format( AttributeError: 'CLIP' object has no attribute 'context_length' > > Any ideas how to override them? Hi, I encountered the same problem as you, how did you solve it, looking forward to your reply. Thanks a lot!

bsakaguchi referenced this issue

2026-10-31 21:13:31 +00:00

[GH-ISSUE #340] RN50x16, RN50x4, ViT-B/32 output nan #294