Tasks 968-969 : Xcopa #358
Conversation
…nstructions-expansion into xcopa � Conflicts: � tasks/README.md
|
@Ravsehajsinghpuri wondering if you can review this PR? |
| { | ||
| "input": "Ese oli mullikilesse mässitud.\n(1)See oli õrn.(2)See oli väike.", | ||
| "output": "1", | ||
| "explanation": "The first alternative makes more sense because you put object is bubble wrap if they are delicate." |
There was a problem hiding this comment.
I think you meant "put object in bubble wrap"
| { | ||
| "input": "Ma tühjendasin oma taskud.\n(1)Ma leidsin pileti tüki.(2)Ma leidsin relva.", | ||
| "output": "1", | ||
| "explanation": "The second alternative makes more sense because finding a ticket in your pocket is more common than finding a gun." |
There was a problem hiding this comment.
The output is "1" but in the explanation you have mentioned "The second alternative makes more sense"
| "explanation": "The second alternative makes more sense because the termites will eat wood." | ||
| } | ||
| ], | ||
| "Negative Examples": [ |
There was a problem hiding this comment.
Can you add one more negative example?
sure! |
|
@amirrezamirzaei Can you update the "Input language" from et to Estonian and similarly for en to English for all the tasks? Similarly for "Output language" and "Instruction language". I think full form of the language is more appropriate. |
|
This PR is good to go when these minor changes are made. Rest all lgtm |
|
This PR lgtm @danyaljj |
|
Please make sure to set the |
|
@danyaljj I'm not sure what the appropriate domain is in these two tasks. the instances don't seem to be about anything particular. |
|
Hi! I think according to the paper, these are made from people's narratives, so I think "Personal Narratives" is a good domain. Please update the tasks in this pr and #386 accordingly. Thanks. |
|
@amirrezamirzaei, can you update the files so that we can go ahead in this PR? |
@amirrezamirzaei reminder to address the reviewer comments please |
|
Could you merge it with the main branch? |
creating two task from Xcopa
they are the same tasks as in copa but in Estonian. this dataset also contains 10 more languages.
submitting this as a test to see whether we want these kinda task or not. each language contains 600 instances. should I make the other language tasks?