Dive into the art of colors in motion graphic videos with our award-winning tool in #chi2024 , Piet! 🏆 Experience our live demo and see how easy and fun tweaking video colors can be! Check it out now: xinyu-shi.github.io/uploads/Piet_p…

If you are interested in colors and AI, you shouldn’t miss Xinyu’s talks tmr at 4pm, one of which is a best paper! #CHI2024 Waterloo HCI Waterloo's Cheriton School of Computer Science

My last work at Microsoft Research is finally released: github.com/microsoft/MS-M… 10 MILLION REAL Bing search queries with 60 MILLON+ REAL user clicks on 10 BILLION ClueWeb22 documents. Have fun scaling up!

Introducing FLAME🔥: Factuality-Aware Alignment for LLMs

We found that the standard alignment process **encourages** hallucination. We hence propose factuality-aware alignment while maintaining the LLM's general instruction-following capability.
arxiv.org/abs/2405.01525

Super honored that Piet get the best paper!! It was so much fun working on this paper and huge thanks to all my awesome co-authors: Yinghou, Yun Wang, and Jian Zhao!

interesting work on long-context embedding. It demonstrate you can effectively extend existing models to a long-context embedding model without any additional training. It also comes with eval data that having a more uniformly distributed target information position in docs.

🧵📢Attention folks working on LONG-document ranking & retrieval! We found evidence of a PROFOUND issue in existing long-document collections, most importantly MS MARCO Documents. It can potentially affect all papers comparing different architectures for long document ranking.⏩

Introducing GRIT🦾to unify text embedding 🔢& generation 📝. GritLM is open SoTA on embedding (MTEB) & generative tasks (BBH etc.) – Both in 1 model. See 🧵for how GRIT🦾 makes RAG >60% faster & more

📜arxiv.org/abs/2402.09906
💻github.com/ContextualAI/g…
1/12

Foundations of Vector Retrieval

This 185-page monograph provides a summary of major algorithmic milestones in the vector retrieval literature, with the goal of serving as a self-contained reference for new and established researchers.

📝arxiv.org/abs/2401.09350

🚀 Introduce UniIR, a unified instruction-guided multimodal retriever handles diverse tasks.
- 1️⃣model for 8️⃣ retrieval tasks (SoTA w/ Instruction-tuning)
- Generalizes to unseen retrieval tasks.
- M-BEIR: multimodal retrieval benchmark w/ 10 datasets, 1.1M queries, 5.6M cands.

We remain committed to our partnership with OpenAI and have confidence in our product roadmap, our ability to continue to innovate with everything we announced at Microsoft Ignite, and in continuing to support our customers and partners. We look forward to getting to know Emmett

Introducing Self-RAG, a new easy-to-train, customizable, and powerful framework for making an LM learn to retrieve, generate, and critique its own outputs and retrieved passages, by using model-predicted reflection tokens.
📜: arxiv.org/abs/2310.11511
🌐: selfrag.github.io

Excited 🤩 to develop Apps using LLMs but puzzled 🤔 over debugging hallucinations?

Thrilled to share AutoDebug, a new transferable way of automated faithfulness testing for LLMs, including self-debug & cross-debug

Arxiv: arxiv.org/pdf/2310.12516…
autodebug-llm.github.io
[1/n]

New entrants into the camelidae family 🦙 for retrieval! Xueguang Ma presents RepLLaMA (a dense retrieval model) and RankLLaMA (a pointwise reranker) fine-tuned on (you guessed it!) LLaMA for multi-stage text retrieval: arxiv.org/abs/2310.08319

New work by Raphael Tang Xinyu Crystina Zhang Xueguang Ma adds yet another prompting technique to the mix: *permutation* self-consistency prompting to overcome positional bias in LLMs. Useful for listwise ranking... read all about it! arxiv.org/abs/2310.07712

🚀 Excited to introduce Positional Skip-wisE (PoSE) training—a revolutionary approach for extending context windows of LLMs to extreme lengths by decoupling train & target length!
Read more in our paper: arxiv.org/abs/2309.10400
Explore the codes: github.com/dwzhu-pku/PoSE

Introducing ChatArena 🏟 - a Python library of multi-agent language game environments that facilitates communication and collaboration between multiple large language models (LLMs)! 🌐🤖

Check out our GitHub repo: github.com/chatarena/chat…

#ChatArena #NLP #AI #LLM 1/8 🧵

CS Professor Jimmy Lin (Jimmy Lin) has been named a 2022 ACM Fellow for his contributions to question answering, information retrieval, and natural language processing. Congratulations on this well-deserved and significant recognition, Jimmy!

cs.uwaterloo.ca/news/jimmy-lin…