short query/ long query/ overall) results

[kaiXuewen]

Hi, Great work!
I was wondering, how do I set the --val_dataset to get the results in Table 1?
The default eval command set it to "ReasonSeg|train", but I don't know how to get the right panel e.g. the test column (short query/ long query/ overall) results.

[rui-qian]

Hi, Great work! I was wondering, how do I set the --val_dataset to get the results in Table 1? The default eval command set it to "ReasonSeg|train", but I don't know how to get the right panel e.g. the test column (short query/ long query/ overall) results.

Apologies for the delayed response. To get the right panel (e.g., the test column results for short query / long query / overall), please uncomment line 184 in the following file:

READ/dataloaders/test_dataset.py

Line 184 in 5ca3250

#for long query

We’re currently working on a systematic framework and will be reorganizing and cleaning up the code. This configuration option will be properly included in the updated version.

Feel free to reach out if you need help!

[rui-qian]

Hi, Great work! I was wondering, how do I set the --val_dataset to get the results in Table 1? The default eval command set it to "ReasonSeg|train", but I don't know how to get the right panel e.g. the test column (short query/ long query/ overall) results.

Apologies for the delayed response. To get the right panel (e.g., the test column results for short query / long query / overall), please uncomment line 184 in the following file:

READ/dataloaders/test_dataset.py

Line 184 in 5ca3250

#for long query
We’re currently working on a systematic framework and will be reorganizing and cleaning up the code. This configuration option will be properly included in the updated version.

Feel free to reach out if you need help!

We recently released a high-performance framework and recommend using it for training: UGround.