HaoKun Lin
@HaokunLin
Followers
6
Following
16
Media
3
Statuses
23
Joint Ph.D. Student at UCAS & CityU
Joined October 2024
🔥 Welcome everyone to our Oral Presentation 'DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs', which will take place in Session 3D: Natural Language Processing, on Thursday, December 12th at 10:20 AM. 🎉. #NeurIPS2024 . 🧵1/5
1
9
10
RT @yshan2u: 🚀 🚀 Introducing ARC-Hunyuan-Video-7B, a model that understands real-world short videos with accuracy and speed. The model exce….
0
11
0
RT @yauchungyiu: Our new quantization-aware training algorithm RoSTE can fine-tune quantized LLMs and break state-of-the-art benchmark accu….
0
2
0
RT @dinq_io: The engine behind Google DeepMind’s Veo 3? . A powerhouse of engineers:.Abhishek Sharma, .Alina Kuznetsova, .Ali Razavi, .Alek….
0
2
0
RT @_June1126: 🔬 The HKU team presents ParallelComp: a training-free technique for efficient context length extrapolation in LLMs—from 8K u….
0
9
0
RT @Yingjia_Wan: Is your model faithfully translating math into formal languages like Lean?.⚖ Introducing "FormalAlign"! #ICLR2025.⁉️To add….
0
27
0
RT @hahahawu2: 💡Unlocking Efficient Long-to-Short LLM Reasoning with Model Merging. We comprehensively study existing model merging methods….
0
11
0
RT @_yixu: 🔥Are we ranking LLMs correctly?🔥. Large Language Models (LLMs) are widely used as automatic judges, but what if their rankings a….
0
33
0
RT @ZhijiangG: 🚀Exciting to see how recent advancements like OpenAI’s O1/O3 & DeepSeek’s R1 are pushing the boundaries! .Check out our late….
0
63
0
RT @chuanyang_jin: How to achieve human-level open-ended machine Theory of Mind?. Introducing #AutoToM: a fully automated and open-ended To….
0
22
0
RT @_akhaliq: RIFLEx. A Free Lunch for Length Extrapolation in Video Diffusion Transformers. TL;DR: Effortlessly extend your video with jus….
0
30
0
RT @haoailab: 🎥 Videos DiTs are painfully slow, HunyuanVideo takes 16 min to generate a 5s 720P video on H100. 🤯. Announcing Sliding Tile….
0
55
0
RT @DAlistarh: Happy to release AQUA-KV!.A new state-of-the-art KV-cache quantization method that achieves near-lossless compression at 2-2….
0
26
0
RT @li_chengzu: Forget just thinking in words. 🚀 New Era of Multimodal Reasoning🚨.🔍 Imagine While Reasoning in Space with MVoT. Multimodal….
0
168
0
RT @YinhongLiu2: 🚨 New Paper Alert! 🚨.When using LLMs for judgements, ever wondered about the consistency of those judgments? 🤔.Check out o….
0
70
0
RT @qiushi_sun: 🎉Introducing our latest work on GUI Agents: "OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synt….
0
42
0
RT @haoailab: 🎥 Frustrated by Sora's credit limits? Still waiting for Veo 2?.🚀 Open-source video DiTs are actually on par. We introduce Fas….
0
62
0
Welcome to our poster at the poster session (East Hall A-C #1911) on Thursday 12 Dec 11 a.m. — 2 p.m.👋.Yichen @Cooper00Cdswf is waiting for you! (He is awesome👀).
0
0
0
RT @ZhijiangG: Excited that HydraLoRA is accepted as Oral #NeurIPS2024! Sad because neither @clin_tian nor I can present it in person. Huge….
0
5
0
🌟 We welcome you to star and use our DuQuant!. 🎤 Project: 📜 Paper: 🔧 Code: 🧵5/5.
github.com
[NeurIPS 2024 Oral🔥] DuQuant: Distributing Outliers via Dual Transformation Makes Stronger Quantized LLMs. - Hsu1023/DuQuant
1
0
0
Extensive evaluations on LLaMA, Vicuna, and Mistral models across PPL, QA, MMLU, MT-Bench, and LongBench benchmarks demonstrate DuQuant's superiority!. For LLaMA2-7B, DuQuant accelerates prefilling phase by up to 2.08× and reduces memory usage in decoding phase by 3.50×. 🧵4/5.
1
0
0