Explore tweets tagged as #WordNet
A new, sophisticated jailbreak technique known as Neural Carrier Articles embeds prohibited queries into benign carrier articles in order to effectively bypass model guardrails. Using only a lexical database like WordNet and composer LLM.
1
0
2
讲个题外话，英语入门后推荐在桌面安装个离线版 en-en 词典，推荐两个开源免费的：wordbook 小巧方便，数据来源于 wordnet。另一个是 kiwix，严格来说它是一个维基数据库浏览器，安装之后用它下载 wiktionary（9GB，可用磁力链下载）就可以象翻维基一样有趣地翻单词了。另：不推荐双语（双向）词典
海关问我为啥来美国 说 travel还不行，说个 sightseeing才放我走…… 我这是来考英语听力的么 😅
5
36
218
소설 쓸 때 유용한 구독 사이트를 소개합니다. (이미 다 아실듯) 단어를 입력하면 유의어와 관련어를 출력해주는 사이트입니다. 한국어 워드넷 사전이에요. https://t.co/iSvcbQzfR9 저는 한글창, 국어사전창, 워드넷 사전창. 이렇게 세 개 켜놓고 작업을 시작해요🤩
1
2K
3K
Hierarchical Retrieval with Dual Encoders • DEs can solve HR if dim. scales with depth/log(# docs) • Challenge: lost-in-the-long-distance — far ancestors hard to retrieve • Solution: Pretrain → finetune on long-distance pairs • Boosts recall on WordNet from 19% → 76%
2
25
176
Machine translators compete in translating the Princeton WordNet concepts to Ukrainian 😀 #unlpworkshop #lrec2024
0
0
7
2024/07/29 ・掲載語彙を大量に追加しました。現在147582の語彙を掲載中です。これにより、WordNet(上位語や下位語 など、意味の繋がりの辞書)の語彙間のリンクが全て機能するようになりました。 ・「もっと覚える」のコラムを100ほど追加しました。 ※イメージ画像や語源の情報は適宜追加予定です。
1
2
24
Using Chat GPT reminds me of being 12 yrs old again. When the coolest thing to do online was talk to your friends on MSN & talk to @UltraHal- a chat bot. Hal won the ‘07 Loebner Prize for "most human" chatterbot (as they used to be called) & utilized wordnet, an early NLP tool.
2
0
11
A nicely presented paper with an interesting message: Categorical concepts in LLMs are represented as simplices. Hierarchically related concepts live in orthogonal subspaces. The findings align with the structure in the WordNet hierarchy.
2
20
169
Petya Osenova is presenting our paper "Recent Developments in BulTreeBank-WordNet" at Global WordNet Conference 2023, San Sebastian, Spain.
0
1
3
An automated way of building the Ukrainian WordNet from Princeton WordNet and WikiData 💡 Find details in https://t.co/zKjNNaojPV
#unlpworkshop #lrec2024
0
0
6
🚀 Exciting research alert! 🚀 GSSR-GNN: A model for constructing geospatial service similarity relationships🌍🔗By leveraging service descriptions, tags, and #BERT & #WordNet, it boosts service replacement efficiency & cuts computational costs🛠️💡 #AI #geospatial #Innovation
0
4
22
Recreated this analysis with Wu-Palmer semantic similarity and NLTK wordnet. Claude and Haiku (via Maestro) suggested I do the analysis this way. Happy with this performance, and charts can be generated on demand.
So I ran the experiment and used a QRNG to find the random word "cur" which is just a mean looking dog. I used embeddings to find the similarity scores. cat: 7.35% hog: 0.61% rat: 2.68% sew: 4.26% run: -5.96% None of these were that significant. However, cat was fairly close
3
0
1
🚀 Filling the gaps in knowledge graphs just got smarter! Felix Hamann, Lukas Walker, & Adrian Ulges, show how LLMs + pre-ranking unlock new entities & boost performance on Freebase, WordNet, and Wikidata #SEMANTiCS2025
4
1
3
Some LLM trainers have been studying ImageNet (WordNet) pretty intensively, some others not so much. The OpenAI one was expected, since it was already explicit part of CLIP's data construction, but I didn't expect it to get every single ID exactly right.
@giffmana tried my best, fed it through resnet and it is confirmed: beach
8
2
62
Day - 11 Didn't do much today as exams are still going on but as these came in exams also , were cool to do > stemming via Porter Stemmer and Regex Stemmer (for fine control) > using lemmatization (higher preference due to easier keyword preservation) via Wordnet
0
0
8
> WordNet 😑
@spreadsheeticus @OhShay481628 @MasinElije An online library of the cumulative different viewpoints of real, living people, not some corporate shill. You didn’t even check it out before you said something. That’s real ignorance.
1
0
3