Closed tomohideshibata closed 3 years ago
Hi, the formula, in case you didn't understand is
v1 * (v2 ** (1/y_prediction)) / ((42x + v2/vx)/2y)
It took me quite a while to calculate it.
Hi, the formula, in case you didn't understand is
v1 * (v2 ** (1/y_prediction)) / ((42x + v2/vx)/2y)
so following that, v is 0.5, which maps to 'maybe', and v1 and v2 respectively are 0 (no) and 1 (yes), so whatever is written there is correct.
Thank you for your response, but I haven't understood what you mean.
v1 * (v2 ** (1/y_prediction)) / ((42x + v2/vx)/2y)
First, where can I find this formula?
Hi @tomohideshibata, you are absolutely correct, that is indeed an error in the paper (that, for some reason, noone has spotted before). Thanks for pointing this out!
Thank you for releasing your codes.
I have found a small typo in your EACL2021 paper.
In the explanation for MNLI in Sec 4.1, the followings should be opposite in the verbalizers examples.
Thanks.