Explore insightful discussions on pivotal AI research and innovations in this podcast hosted by Arize AI founders. Discover the minds and methods shaping machine learning.
Accurate KV Cache Quantization with Outlier Tokens Tracing
25 mins • Jun 4, 2025
Charts
- 18Decreased by 2
'/%3e%3cpath%20d='M74.965,77.269%20C74.795,75.733%2074.283,74.624%2073.259,73.6%20C71.339,71.595%2067.968,70.272%2064,70.272%20C60.032,70.272%2056.661,71.552%2054.741,73.6%20C53.76,74.667%2053.205,75.733%2053.035,77.269%20C52.693,80.256%2052.907,82.816%2053.248,86.955%20C53.589,90.88%2054.229,96.128%2055.04,101.419%20C55.637,105.216%2056.107,107.264%2056.533,108.715%20C57.259,111.104%2059.861,113.152%2064,113.152%20C68.139,113.152%2070.784,111.061%2071.467,108.715%20C71.893,107.264%2072.363,105.216%2072.96,101.419%20C73.771,96.085%2074.411,90.88%2074.752,86.955%20C75.136,82.816%2075.307,80.256%2074.965,77.269%20z'%20fill='%23FFFFFF'%20id='XMLID_657_'/%3e%3cpath%20d='M74.496,55.509%20C74.496,61.312%2069.803,66.005%2064,66.005%20C58.197,66.005%2053.504,61.312%2053.504,55.509%20C53.504,49.707%2058.197,45.013%2064,45.013%20C69.803,45.013%2074.496,49.749%2074.496,55.509%20z'%20fill='%23FFFFFF'%20id='XMLID_655_'/%3e%3cpath%20d='M63.872,14.165%20C39.381,14.251%2019.328,34.133%2018.987,58.624%20C18.731,78.464%2031.403,95.445%2049.109,101.675%20C49.536,101.845%2049.963,101.461%2049.92,101.035%20C49.707,99.499%2049.451,97.963%2049.28,96.427%20C49.195,95.872%2048.853,95.445%2048.384,95.189%20C34.389,89.088%2024.619,75.051%2024.789,58.795%20C25.003,37.461%2042.411,20.096%2063.701,19.925%20C85.504,19.755%20103.296,37.419%20103.296,59.179%20C103.296,75.264%2093.568,89.088%2079.701,95.189%20C79.189,95.403%2078.848,95.872%2078.805,96.427%20C78.592,97.963%2078.379,99.499%2078.165,101.035%20C78.08,101.504%2078.549,101.845%2078.976,101.675%20C96.512,95.531%20109.099,78.805%20109.099,59.179%20C109.013,34.347%2088.747,14.123%2063.872,14.165%20z'%20fill='%23FFFFFF'%20id='XMLID_653_'/%3e%3cpath%20d='M62.848,29.099%20C46.891,29.696%2034.091,42.88%2033.877,58.837%20C33.749,69.333%2038.997,78.635%2047.061,84.139%20C47.445,84.395%2048,84.096%2048,83.627%20C47.872,81.792%2047.872,80.171%2047.957,78.464%20C48,77.909%2047.787,77.397%2047.36,77.013%20C42.453,72.405%2039.467,65.835%2039.637,58.581%20C39.979,45.781%2050.261,35.371%2063.061,34.859%20C76.971,34.304%2088.405,45.483%2088.405,59.221%20C88.405,66.219%2085.419,72.533%2080.683,77.013%20C80.299,77.397%2080.085,77.909%2080.085,78.464%20C80.171,80.128%2080.128,81.749%2080.043,83.584%20C80,84.053%2080.555,84.395%2080.981,84.096%20C88.917,78.677%2094.165,69.504%2094.165,59.179%20C94.208,42.197%2080,28.416%2062.848,29.099%20z'%20fill='%23FFFFFF'%20id='XMLID_651_'/%3e%3c/g%3e%3c/svg%3e)
- 21Increased by 0
- 25Decreased by 2
- 17Increased by 13
- 34Decreased by 1
Recent Episodes
Jun 4, 2025
Accurate KV Cache Quantization with Outlier Tokens Tracing
25 mins
May 16, 2025
Scalable Chain of Thoughts via Elastic Reasoning
29 mins
May 2, 2025
Sleep-time Compute: Beyond Inference Scaling at Test-time
30 mins
Apr 18, 2025
LibreEval: The Largest Open Source Benchmark for RAG Hallucination Detection
27 mins
Apr 4, 2025
AI Benchmark Deep Dive: Gemini 2.5 and Humanity's Last Exam
26 mins
Language
English
Country
United States
Feed Host
Website
Feed
Request an Update
Updates may take a few minutes.