Dataset
#1
by
infinitylogesh
- opened
Thank you for this model ! Can you please help with the details of the dataset used to train this ? Was it - theblackcat102/evol-codealpaca-v1 as mentioned in the github link ? Thanks.
20k from there and 20k is new that was generated using synthetic methods for repair and explanation
Thank you for the response. This model shows a very good improvement to Humaneval from the base model. I am curious and would be great , If you can share some details of how the 20k was filtered from evol-codealpaca-v1 and some details of the new synthetic dataset that was created ? Thanks in advance.
I'm also curious about your data filtering method here? Also, which benchmark is your model test score?