✨ Still not satisfied with your speculative decoding (SD)? ✨ Try our newly released ParallelSpec 🚀, which integrates seamlessly into existing SD frameworks! 💡 By carefully training a parallel drafter, ParallelSpec can generate future tokens in parallel, boosting leading SD

🚀Introducing OpenWebVoyager: a multi-modal, LLM-based web agent built on open-source models! It iteratively improves through real-world exploration, followed by visual critique and optimization using successful trajectories after each round! Paper: arxiv.org/abs/2410.19609 Code:

🥳 We open-sourced Leopard-Instruct, a dataset containing 𝟏𝐌 𝐡𝐢𝐠𝐡-𝐪𝐮𝐚𝐥𝐢𝐭𝐲, 𝐭𝐞𝐱𝐭-𝐫𝐢𝐜𝐡, 𝐦𝐮𝐥𝐭𝐢-𝐢𝐦𝐚𝐠𝐞 𝐢𝐧𝐬𝐭𝐫𝐮𝐜𝐭𝐢𝐨𝐧-𝐭𝐮𝐧𝐢𝐧𝐠 examples. It significantly improves performance on multi-image understanding! Github: github.com/tencent-ailab/…

Join us at the Workshop on Bridging Neurons and Symbols for Natural Language Processing and Knowledge Graphs Reasoning COLING 2025 on January 20, 2025. Submission deadline: Deadline: December 1, 2024 neusymbridge.github.io/call-for-paper…

Excited to revisit Vancouver😆 Our team at Tencent AI Lab is actively hiring full-time researchers and research interns for next year. Let's chat if you are interested in conducting frontier research in Multi-modal agents, RL, and model architectures.😼😼😼

Announcement #1: our call for papers is up! 🎉 colmweb.org/cfp.html And excited to announce the COLM 2025 program chairs Yoav Artzi Eunsol Choi Ranjay Krishna and Aditi Raghunathan

Thrilled to share that my PhD student, Nan Jiang, is on the academic job market! Nan, working on SE+AI, is the ideal kind of PhD student one can ask for. I’m confident he will be a great professor! jiang719.github.io Nan has papers in ICSE, FSE, NeurIPS, AAAI, and more.

Excited to see Genesis. Have been looking for this for a long time. It integrates ultra-fast differentiable physics with a generative framework, making it capable of training agents in dynamic, realistic environments. Being differentiable is so cool!!!

We have moved the code repo to github.com/Tencent/CogKer…. 😀

Thrilled to see Google’s Project Mariner hit 83.5% on our WebVoyager benchmark for real-world web tasks! Excited for more open-source efforts. Explore our benchmark and code here: github.com/MinorJerry/Web…

Excited to see our LongMemEval featured by LongBench v2! Tackling long-context challenges in real-world scenarios is key to enabling LLMs to retain user histories and preferences. Can’t wait to see it spark innovation in long-context and memory mechanisms! Also directly checkout

🤖 Tired of slow tree searches on LLMs? 🚀 Check out our latest research on efficient tree search! 🔹 We introduce an upgraded transformer architecture that enables token-level self-reward modeling (TRM). 🔹 On top of that, we developed the Streaming Looking Ahead (SLA)

🎉Excited to share that our system got accepted to NAACL 2025! Looking forward to meeting everyone in Albuquerque soon! 🌵✨ #NAACL2025 #AI #NLP #MachineLearning

Attending NAACL and will host the KnowledgeNLP workshop on Saturday! Happy to chat about LLM agent, multimodal, and reasoning and RAG topics! ☕️

We will be demonstrating our system at #NAACL25 2 PM today. Feel free to stop by and try it😆

🚀 Check out our paper: WebEvolver: Enhancing Web Agent Self-Improvement with Coevolving World Model, from Tencent AI Lab!. We present a world model-driven framework for self-improving web agents, addressing critical challenges in self-training—such as limited exploration and