Monday-Thursday

Revising and submiting my master thesis

[x] Done

Friday

Start doing my research.

First, headup to my teammates.

[x]Teammate meeting

[x]Yufeng

[x]Runpeng

[x]Read Samson’s paper

Slides can be found at: https://www.comp.nus.edu.sg/~kanmy/talks/Its%20Morphin%20Time%20Combating%20Linguistic%20Discrimination%20with%20Inflectional%20Perturbations.pdf

To summary, this paper discovers that morphological inflection is a discrimination!

• Many models cannot perform well given those morphological inflections — discover that it is a flaw for many state-of-the-art models.
• Fine-tuning existing models can better the performance.
• • I think it is a plus, because it provides solutions.

[]Submit our KDD work to ArXiv

[]We get a format error to fix first.

[] Short term: Intro Done for our AAAI submission

[]Ask Yunshan for the teamup.com usage of DGX GPUs.

[x]Found a very stable way to connect to DGX servers.

[x]Installed conda and torch.

[] Long-term: Study shortcuts in terminals… How to directly login to a server, How to open my website git repo, etc…

Reading Chris Manning’s paper

• Can we have a definition on self supervision?
• • Self-supervised learning allows us to learn good representations without using large annotated databases. Instead, we can use unlabeled data (which is abundant) and optimize pre-defined pretext tasks. We can then use these features to learn new tasks in which data is scarce.

Reading wolf’s log:

He gives a bold question:

https://medium.com/huggingface/learning-meaning-in-natural-language-processing-the-semantics-mega-thread-9c0332dfe28e

Saturday

[] Short-term: REWRITE the Intro part of our AAAI submission.

[] Short-term: add image / icon to my website!