📣 Excited to announce that we contributed to Aurora-M, the first #multilingual, #redteamed and #opensource #LLM! Kudos to @SimoneTedeschi_ @RNavigli, and all other authors! 📄 Read the paper here: lnkd.in/dMqxK2MV 🧵Follow the thread for a summary!

Thrilled to announce that we received the *outstanding paper award* at ACL 2023 for our paper "What's the meaning of superhuman performance in today's NLU"! babelscape.com/research#pubs #NLProc #ACL2023 #ACL2023NLP

We are presenting 6 research papers on a variety of #NLProc topics (#LLMs, #summarization, #RelationExtraction, #SemanticParsing and alignment) at #ACL2023! Don't miss them! babelscape.com/research#pubs #ACL2023NLP #NLU

📄 Paper: arxiv.org/abs/2306.13467 🚀 Github: github.com/SapienzaNLP/LeakD… #AMRParsing #ACL2023NLP #ResearchPaper 🧵5/5

💡 Through self-knowledge distillation and word-to-node alignment, we achieve state-of-the-art AMR parsing results. Our experiments demonstrate the effectiveness of our approach. #StateOfTheArt #NLPResearch 🧵4/5

Check out our new paper at #ACL2023 on the integration of graph information in Transformer-based semantic parsing.📄 By @Pavlo_Vasylenko, @PereLluisHC, @CarlosMalaga26, @RNavigli #ACL2023NLP #NLProc #AMRParsing @knowgraphs 🧵1/5

Check out our new #ACL2023 paper on overcoming current limits of ensembling semantic parsing models! 📄arxiv.org/abs/2306.10786 @CarlosMalaga26 @PereLluisHC and @RNavigli #ACL2023NLP #NLProc #AMRParsing #semanticparsing @knowgraphs 🧵1/5

Ecstatic to share exciting news about a fresh business collaboration, with @OxLanguages, the #1 globally recognized brand leader in #multilingual #lexical research & publishing! More here: babelscape.com/#news #dictionaries #nlp #nlproc #lexicography #multilinguality

📄 Paper: arxiv.org/abs/2306.04334GitH…: 🚀Github: github.com/Babelscape/echoes…… #ACL2023 #BookSummarization #EchoesFromAlexandria #MultilingualNLP 🧵6/6

🏆When evaluated on BookSum, our LED-based approach achieves state-of-the-art performance (ROUGE-1: 42.13, ROUGE-2: 10.53, and ROUGE-L: 16.75). Our achievement is attained using only 0.1% of the parameters utilized by the GPT-3-based competitor (Wu et al., 2021). 🧵5/6

📣Check out our new #ACL2023 paper! "Echoes from Alexandria: A Large Resource for Multilingual Book Summarization" 📄arxiv.org/abs/2306.04334 @alescire94 @ConiaSimone SimoneCiciliano and @RNavigli #ACL2023NLP #NLProc #BookSummarization #MultilingualNLP @ERC_Research 🧵 1/6