I have encountered several challenges while attempting to replicate your experiment, and I am in urgent need of your assistance. about learning-to-defer-with-limited-expert-predictions

[hechengbo-H]

The error was concentrated in this part.

[hechengbo-H]

[hechengbo-H]

hi，I attempted to write the correct tags and ran the code. However, I only executed the embedding_cm as per the readme instructions, which led me to the directory learning-to-defer-with-limited-expert-predictions-main/Learning-to-Defer-Algs/mozannar_sontag. Although I ran the code, the results that I obtained do not seem to match those reported in the paper. I am uncertain whether this is due to a problem with my execution or a flaw in the code itself. Could you please provide me with the complete code for running the algorithm? My email is [email protected]

[lukasthede]

Hello,
for generating the synthetic expert labels for the CIFAR100 experiments you can refer to Experiments section, subsection CIFAR100/Dataset and Expert Generation.

We have also updated the repository to include the code for generating the synthetic CIFAR100 expert labels in the directory /Synthetic-CIFAR-Expert. Using this code you can generate the TrueEx_cifar100_60_labels.json file and use it for training the learning to defer algorithms.

[hechengbo-H]

Hi, thanks for your reply, I will keep trying to repeat your experiment, thank you.