Using truncation in idefics3 processor

#1
by jainarchit76 - opened

Is it possible to use truncation with processor(text=texts, images=images, return_tensors="pt", padding=True, truncation=True, max_length=512).
When using max_length of 512, it raises an error:
RuntimeError: shape mismatch: value tensor of shape [2197, 4096] cannot be broadcast to indexing result of shape [485, 4096]

So? What is the point of actually using the max_length truncation parameter?

Sign up or log in to comment