redditPhD in ML in the AI era, autoresearch [D]redditmachine-learningMachineLearning4 days ago▲ 10Read story→
redditLARA: small, composable behaviours for frozen LLMs [P]redditmachine-learningMachineLearning4 days ago▲ 10Read story→
redditTabPFN-3.5 is released as the next SOTA tabular foundation model [N]redditmachine-learningMachineLearning5 days ago▲ 10Read story→
redditNeurIPS 2026: handling of multiple venue locations seems bad [D]redditmachine-learningMachineLearning5 days ago▲ 10Read story→
YhackernewsShow HN: Ordewell – turn one goal into an ordered plan of coding-agent taskshackernews5 days ago▲ 10Read story→
redditI trained a 44M parameter quantized LLM from scratch on 45B tokens. It ships in 19.8 MB and runs at ~1,900 tok/s on CPU. [P]redditmachine-learningMachineLearning5 days ago▲ 10Read story→
YhackernewsDiplodocus, Long Thought Exclusively American, Turns Up in Spainhackernews5 days ago▲ 10Read story→
redditHow much work in progress can a workshop submission be [R]redditmachine-learningMachineLearning5 days ago▲ 10Read story→
reddit[D] How do you get preprocessed dataset of a paper [D]redditmachine-learningMachineLearning5 days ago▲ 10Read story→
YhackernewsDestroy After Reading: photocopiers,cheap paper and DIY gave metal it's lookhackernews6 days ago▲ 10Read story→
redditHow to automatically find the batch size when using Accelerate with FSDP2? [D]redditmachine-learningMachineLearning6 days ago▲ 10Read story→
YhackernewsCloudflare AKE cuts origin HelloRetryRequests from 52% to 3.7%hackernews6 days ago▲ 10Read story→
redditMS MARCO click-translation expansion tables ("poor man's" DSSM) [P]redditmachine-learningMachineLearning6 days ago▲ 10Read story→
redditDuplicating baseline benchmarks [D]redditmachine-learningMachineLearning6 days ago▲ 10Read story→
redditSHADOW 50M: a 19.8 MB model that computes exactly and remembers from disk [P]redditmachine-learningMachineLearning6 days ago▲ 10Read story→