Spaces:
Running
on
Zero
Running
on
Zero
how the handle hallucination?
#2
by
RaushanA
- opened
Hi @RaushanA
I suggest that you ask the model one question at a time. This will increase the model's accuracy. If your questions require some math operations, I suggest you format your input like "program of thought: your_question". This will give you a python script that can be executed to calculate and print the final answer.
Regarding finetuning, you can find a finetuning colab notebook with good instructions in the gihtub repo: https://github.com/vis-nlp/ChartGemma
ahmed-masry
changed discussion status to
closed