with the current test dataset, Task B scores cannot be computed. Execution fails with the following error:
File "scoring.py", line 26, in <lambda>
lambda row: row["anchor_embedding"].dot(row["a_embedding"]), axis=1
AttributeError: ("'float' object has no attribute 'dot'", 'occurred at index 0')

Task B is entirely contingent on Task A outputs. The scoring logic for Task B requires exact 1:1 matchingof each text entry from Task B to its corresponding output in Task A. This dependency is verifiable in both the training and development datasets.

Could you please confirm whether these discrepancies are intentional or if they indicate a data issue?
If this is a data problem, we kindly suggest:
Providing a corrected dataset​ to ensure consistent embedding calculations

Dear Organizers,
I have identified several issues with the data provided during executing the following calculation:
Only 55 entries​ of anchor_text matched
254 entries​ of text_afrom File A matched
229 entries​ of text_bfrom File A matched

Hello, I would like to know when the test phase data will be released? I saw on the homepage that the labels for the test data can only be published after the shared task is completed. But where is the unlabeled test data? Have I not found it, or has the organizer not released the data yet?