3 Photos and videos
tj-coding retweeted
Apr 10
Many people (myself included) are wondering:
If not LLMs, what should I work on?
In today's blog, I provide a unified framework for finding blue-ocean opportunities in AI, especially in this chaotic era of LLMs.
kindxiaoming.github.io/blog/…
28
107
899
73,120
Mar 21
🫎 We're pleased to announce that our paper, "Cosine Similarity as Logits", is accepted to the #EACL2026 main conference, Morocco 🇲🇦.
I'll be presenting: 🦌DEcoder Embedding based Relational Probe🦌, a knowledge probe for generative LLMs with effectively linear scaling order.
2
2
209
Mar 21
1
84
Mar 21
GitHub - TJ-coding/deer: DEcoder-Embedding Based Relational KGC / Probe (DEER) · GitHub share.google/5vzD4bqvEam5qY2…
67
tj-coding retweeted
8 Mar 2025
審査なし発表（研究会・全国大会発表）が審査付き発表（国際会議・ジャーナル）に繋がってるのかどうかも気になるのよね。
8 Mar 2025
Science of Scienceに興味を持ってるので、研究コミュニティ形成のメカニズムを明らかにしたいですよね！！
1
4
1,147
2 Mar 2025
左：私の手法、右：新手法
マルチドロップレットトラッキングに強くなっているらしいです.ノイズにも強くなっている様に見えます.【感動的です】
今後も改善を続けて、価値のあるものになればと思います.
レポート：drive.google.com/file/d/11-Q…
73
tj-coding retweeted
19 Nov 2024
New Anthropic research: Adding Error Bars to Evals.
AI model evaluations don’t usually include statistics or uncertainty. We think they should.
Read the blog post here: anthropic.com/research/stati…
50
297
2,087
756,249
tj-coding retweeted
Thrilled to announce that our paper "Artificial intelligence for literature reviews: opportunities and challenges" has been finally published by the 𝘈𝘳𝘵𝘪𝘧𝘪𝘤𝘪𝘢𝘭 𝘐𝘯𝘵𝘦𝘭𝘭𝘪𝘨𝘦𝘯𝘤𝘦 𝘙𝘦𝘷𝘪𝘦𝘸 journal.
#AI #SLR #research #openaccess #LLMs
📰doi.org/10.1007/s10462-024-1…
1
9
18
898
tj-coding retweeted
20 Aug 2024
pubs.acs.org/doi/full/10.102…
磁性材料に特化したBERTモデル「MagBERT」を開発し、Curie温度の自動抽出に成功。文献データの効率的な解析と材料科学のデータ不足解消に貢献。様々な物性や分野への応用に期待。
#マテリアルズ・インフォマティクス
8
33
3,175
tj-coding retweeted
5 Jun 2024
Are LLMs able to understand observational data, derive the underlying rules that govern it, and use these rules to better model new data instances?
In our new paper, @_emliu investigates LMs' ability to perform this *abductive reasoning* from a number of different perspectives.
Abductive reasoning is the ability to propose best explanations for observed data. In our paper we ask: how well can LMs do this by proposing rules that explain data, and how is this ability correlated with other abilities such as in-context learning?
arxiv.org/abs/2404.03028 🧵
13
57
9,922
tj-coding retweeted
16 Mar 2024
Quite an Interesting analysis by Jennifer Hsia, Afreen Shaikh, @ZhiruoW & @gneubig : arxiv.org/abs/2403.09040
Awaiting the RAGGED repo to be published: github.com/neulab/ragged
2
11
53
6,102