im using 2.0 flash for latex rendering, for my task using api, the task is to return a latex for rendering the text from the extracted text in the image, but the model is returning bunch of \n till the token limit, it is not happening all the time but for the images with large text (20 lines), is this an known issue, anybody facing the same?
@Hari_Jaybal
welcome to the community,
I have tried to replicate this issue to generate latex code from image/screen shot of a recently published paper. I am not able to reproduce this ,
is this a persistent issue ? how often does this happen ?
could you please help us with more details like if you were using the model in AI studio or gemini api , also if there are any images or tables in the uploaded images? please give us any details that would help replicate this issue
Thank you
So the issue is not latex generation, the issue is that my prompt is designed to handle different types of question image and extracting the text, i cannot fill in with the detail but the issue is presistant if there is a table in the question and the table is big, im using api by the way not ai studio or web interface
What temperature are you using? I’d recommend setting to 0 for this task.
Im using 0.3, yeah ill try that