google/gemma-3-27b-it running into token limit

#719
by auerovkh - opened

When using google/gemma-3-27b-it, I've noticed I run into the error

Input validation error: inputs tokens + max_new_tokens must be <= 4096. Given: 4182 inputs tokens and 0 max_new_tokens

After 3-4 messages. Should/can this be increased?

Ive noticed this too. Is there no way to trim older data as the conversation grows to make room? None of the other models have this issue.

Sign up or log in to comment