Truncation does not happen with CLIP models.

[GChrysostomou]

Was trying out CLIP via sentence-transformers however I have an error following tokenization that the input has not been truncated. Not sure if this is down to the transformer's version so attaching all information below.

Error:

RuntimeError: The size of tensor a (129) must match the size of tensor b (77) at non-singleton dimension 1

Packages:

sentence-transformers = 3.4.0
transformers = 4.48.1

This refers to this line ->

sentence-transformers/sentence_transformers/models/CLIPModel.py

Line 71 in a624f0c

encoding = self.processor.tokenizer(texts_values, return_tensors="pt", padding=padding)

where to work I needed to add:

..padding, truncation = True)