CompressionBench asks AI agents to shrink language models and build working inference code. 29 of 36 submissions passed verification. Building smaller models is becoming an agent benchmark.
Source:
#AIResearch #ModelCompression
0007
Post con #ModelCompression
#ModelCompression
Post pubblici su X che usano #ModelCompression.
Hashtag correlati
cyber genius@cyber_spanishAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. By @antgrasso #AI #ModelCompression #Efficiency
124228
Visual-Intelligence@VI_Journal_CSIG🔥Researchers from Beihang University and ETH Zurich conducted a systematic evaluation of Qwen3's robustness under various quantization settings. Check out the paper at: @qin_haotong #Quantization #LLM #Modelcompression00289
🎉 Thrilled to share that our paper "Surprising effectiveness of pretraining ternary language models at scale" earned a spotlight at #ICLR2024! We dive into Ternary Language Models (TriLMs), systematically studying their training feasibility and scaling laws against FloatLMs. 1/5
3126312.4K- @imtejas13Performance highlight: At 3.9B parameters, TriLMs rival FloatLMs across reasoning, knowledge, and toxicity benchmarks—despite being 5.9× smaller in bit size! 4/5
105374 - @imtejas13015299
AI GURU@TheHighway2AIAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. rt @antgrasso #AI #ModelCompression #Efficiency
00352
Technology and Businesses@TechnoNBusinessAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. By @antgrasso #AI #ModelCompression #Efficiency
01673
Fraunhofer IKS@FraunhoferIKSPaper presentation: Our colleague presented his new paper “Beyond Test Accuracy: The Effects of Model Compression on CNNs” at this year´s SAFE AI. Watch his presentation here: ➡️ #ai #modelcompression #deeplearning #research #SAFEAI
014
Felice Ragone@felice_ragone78AI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. Microblog by @antgrasso #AI #ModelCompression #Efficiency
00355
Bella Byte@RecipeGrowAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. rt @antgrasso #AI #ModelCompression #Efficiency
03696
Meetween@MeetweenEUMeetween is part of the organising committee of #IWSLT2026 — the premier conference on spoken language translation. The Shared Task Evaluation Period is open. Working on #speechtranslation, instruction following, or #modelcompression? Get involved now! 🔗
044144
Digital SMEs Goals@SmesGoalsAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. rt @antgrasso #AI #ModelCompression #Efficiency
11689
STEM & YOU@you_stemAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. rt @antgrasso #AI #ModelCompression #Efficiency
01382
Chi-Heng Lin 貓貓🐱@ChiHengHenry🚀 ICLR 2025 Oral Presentation! 🚀 Exciting to see our work on modular LLM compression has been selected for an Oral at #ICLR2025! 🎉🎉 If you missed it, check out our key contribution: ⬇️⬇️ #LLM #ModelCompression #MatrixDecompositionChi-Heng Lin 貓貓🐱 @ChiHengHenry🔥🔥Thrilled to share our work @iclr_conf🔥🔥 🤔 Low-rank decomposition for LLM compression adds significant overhead by splitting 1 matrix into 2. 🔑 Key contribution: ✅ Introduction of decomposition-compression duality: By leveraging the duality, we compress two adjoint matrices within a module, eliminating parameter overhead and achieving state-of-the-art performance! 🚀: w/@jamessealesmith @TuliShikhar 📜:228661- Chi-Heng Lin 貓貓🐱@ChiHengHenry🔥🔥Thrilled to share our work @iclr_conf🔥🔥 🤔 Low-rank decomposition for LLM compression adds significant overhead by splitting 1 matrix into 2. 🔑 Key contribution: ✅ Introduction of decomposition-compression duality: By leveraging the duality, we compress two adjoint matrices within a module, eliminating parameter overhead and achieving state-of-the-art performance! 🚀: w/@jamessealesmith @TuliShikhar 📜:
13213.1K 
Social Tech Mantra@SocialTMantraAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. rt @antgrasso #AI #ModelCompression #Efficiency
105105
C-Suite Tech Point@CsuiteTechPointAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. By @antgrasso #AI #ModelCompression #Efficiency
03780
Tamer@TamerGhattas911Excited to share: our paper “On Pruning State-Space LLMs” was accepted to EMNLP 2025! 🎉 Preprint: Code: Model: Smol-Mamba-1.9B → w/ @MichaelHassid & @royschwartzNLP (HUJI) #Mamba #ModelCompression
2414779
Shrink It, Train It Small, or Search for It: The Main Strategies for Small Models, Compared A capable small model can be reached by compressing a large one, training a small one on purpose, searching… Read more: #SmallModels #ModelCompression00085
Maximum Digital Sight@MaxDigitalSightAI model compression isn't just a technical refinement but a strategic choice that aligns cost reduction, sustainability, and operational agility with the pressing demands of today's rapidly evolving digital landscape. By @antgrasso #AI #ModelCompression #Efficiency
01557
Accuracy hides issues in quantized LLMs. 'Accuracy is Not All You Need' (NeurIPS 2024) shows 'flips' & KL-Divergence reveal output changes. #AI #ModelCompression #LLM
113341
📢 Our paper "Quantifying Knowledge Distillation Using Partial Information Decomposition" will be presented at #AISTATS2025 on May 5th, Poster Session 3! Our work brings together #modelcompression and #explainability through the lens of #informationtheory Link:
15181.8K- Mostra altri post
