🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

"I'm in a cheap hotel in California which doesn't have a good internet or phone connection. I was going to have an MRI scan today but I'll have to cancel that!" - New physics laureate Geoffrey Hinton speaking at today’s press conference where his #NobelPrize was announced.

🥰Thanks for the kind words and for helping to promote our work MT Group at FBK Andrea Piergentili. As Andrea Piergentili mentioned, using LLMs as evaluators is an exciting and interesting direction, enabling simpler automatic post-editing pipelines. Those interested in fine-grained MT evaluation and APE are

We wrapped up CS 8803 "Large Language Model" class at Georgia Tech for Fall 2024. Here is the reading list: • learning from human preferences (PPO, DPO, SimPO, CPO, RRHF, ORPO, CTO) • real-world LLM (Llama-3, Aya, Arena's) • efficient LLM (MoMa, LoRA, QLoRA, LESS)

Cool work

Can’t agree more with the opinion that ChatGPT was an UX and HCI driven research, I have same thoughts since the early of last year🤔

Congrats to Wenbin! Welcome to check out our new work ✨HR-Bench accepted by AAAI 2025, where we released the first multimodal high-resolution data set (containing both 4K & 8K) for the multimodal LLM community, also, we presented a training-free framework that enhances

It’s obvious racial bias, there should be no place for it in academia and society😡

How far is an LLM from not only understanding but also generating visually? Not very far! Introducing MetaMorph---a multimodal understanding and generation model. In MetaMorph, understanding and generation benefit each other. Very moderate generation data is needed to elicit

Excited to share that Qingyu will be presenting our paper, ‘MQM-APE: Toward High-Quality Error Annotation Predictors with Automatic Post-Editing in LLM Translation Evaluators,’ at COLING 2025. Don’t miss it on Jan 23 at 11:15 in Capital Suite 7, ADNEC, Abu Dhabi. Come and check

Just had a quick look at DeepSeek's new Janus Pro paper. I don't think it's a big deal (yet...!), but quick TL;DR below before hype gets out of hands.

Whitespace-ignoring tokenization is an fundamental feature of Sentenepiece, implemented since its early stages (around 2017) Using whitespace yielded better results on MT. It would be helpful if you could mention this. github.com/google/sentenc…

Happy to see two of our papers are accepted by #ICML2025, with one spotlight😀 congrats to all students and coauthors. 1/2) Retrieval-Augmented Perception: High-resolution Image Perception Meets Visual RAG. TL;DR: We propose Retrieval-Augmented Perception (RAP), a training-free

Welcome to check out our new work on the efficient LLM Agent - Agent EarlyExit. Congrats to Qingyu-Lu👏 Btw, also welcome to read our other efficient agent work, AgentDropout (arxiv.org/abs/2503.18891 accepted to ACL'25).

Thrilled to announce that the work “Retrieval-Augmented Perception: High-resolution Image Perception Meets Visual RAG” has been selected as the 🎉oral presentation🎉 at #ICML #ICML2025