Inquiry Regarding Experiment with Aristo-RoBERTa Encoder on OBQA Dataset in GreaseLM Paper

[EchoDreamer]

I am reaching out to seek assistance regarding my attempts to reproduce the experimental results mentioned in the GreaseLM paper, specifically concerning the utilization of the Aristo-RoBERTa encoder on the OBQA dataset.

Despite multiple attempts, I have been unable to replicate the performance reported in the paper. In order to facilitate my efforts, I would greatly appreciate it if you could provide more comprehensive details regarding the hyperparameters used in this particular experiment.

Your guidance on this matter would be immensely valuable to me, and I am eager to hear from you at your earliest convenience. Thank you very much for your attention to this matter.

[chit-ang]

Have you succeeded in replicating it so far

[EchoDreamer]

Not yet. Do you have an opinion on this?

[chit-ang]

Could you please reproduce the results on roberta-large

[chit-ang]

I was unable to load the model directly from the transforms while replicating roberta-large. I downloaded roberta-large under FacebookAI in huggingface, but the training result was very poor, only more than 30%, have you ever encountered this problem?

[EchoDreamer]

I was unable to load the model directly from the transforms while replicating roberta-large. I downloaded roberta-large under FacebookAI in huggingface, but the training result was very poor, only more than 30%, have you ever encountered this problem?

I downloaded the roberta-large model directly from the huggingface mirror site（https://hf-mirror.com/） and then trained it, and the results were close to the original paper's results

[chit-ang]

不知道是不是方便加个联系方式，我最近也在复现这篇文章可以一起交流一下

[biegekekeke]

不知道是不是方便加个联系方式，我最近也在复现这篇文章可以一起交流一下

请问复现这篇论文需要多大的显存？