Hacker News
- Monitor and Optimize your large-scale model training https://trainy.ai/blog/introducing-trainy 2 comments
- Why train when you can optimize? https://justinmeiners.github.io/why-train-when-you-can-optimize/ 68 comments
- Show HN: AdaBound, an optimizer that trains as fast as Adam and as good as SGD https://github.com/Luolc/AdaBound 6 comments
- Sophia: Scalable Stochastic 2nd-Order Optimizer for Language Model Pre-Training https://arxiv.org/abs/2305.14342 2 comments
- Training ResNet-50 on ImageNet in 35 Epochs Using Second-Order Optimization https://arxiv.org/abs/1811.12019 10 comments
- Behold my insanity! Buffered & beaconed train-to-train foundry (hopefully UPS optimized) https://www.reddit.com/r/factorio/comments/oakatt/behold_my_insanity_buffered_beaconed_traintotrain/ 13 comments factorio
- What's the best "extra-optimal" train unloader? https://www.reddit.com/r/factorio/comments/jcdguu/whats_the_best_extraoptimal_train_unloader/ 16 comments factorio
- New Google Brain Optimizer Reduces BERT Pre-Training Time From Days to Minutes https://medium.com/syncedreview/new-google-brain-optimizer-reduces-bert-pre-training-time-from-days-to-minutes-b454e54eda1d 8 comments artificial
- New Google Brain Optimizer Reduces BERT Pre-Training Time From Days to Minutes https://medium.com/syncedreview/new-google-brain-optimizer-reduces-bert-pre-training-time-from-days-to-minutes-b454e54eda1d 5 comments deeplearning
- New Google Brain Optimizer Reduces BERT Pre-Training Time From Days to Minutes https://medium.com/syncedreview/new-google-brain-optimizer-reduces-bert-pre-training-time-from-days-to-minutes-b454e54eda1d 4 comments learnmachinelearning
- [R] Numerical Instability in Some Optimizers for training Neural Network https://arxiv.org/abs/2307.16189 4 comments machinelearning
- Open-source optimization framework for deep learning models (AI) training could cut the carbon footprint by up to 75% https://news.umich.edu/optimization-could-cut-the-carbon-footprint-of-ai-training-by-up-to-75/ 5 comments science
- Optimizing 2 Way Trains https://factorioblueprints.tech/blueprint/85881a70-68e0-4a92-9720-87c10ac92bac?selected=e415b42f-5797-4519-aaaf-1c04129a811c 7 comments factorio
- Going megabase: Request for train advice & UPS optimization https://www.reddit.com/r/factorio/comments/o8xqpw/going_megabase_request_for_train_advice_ups/ 8 comments factorio
- A comprehensive guide to understanding how optimizers work in training neural networks. Explains Momentum, Nesterov, Adagrad, RMSProp and Adam. https://medium.com/deep-learning-demystified/https-medium-com-deep-learning-demystified-understanding-optimizers-313b787a69fe 24 comments learnmachinelearning
- How do you optimize your trains? https://www.reddit.com/r/factorio/comments/949nr0/how_do_you_optimize_your_trains/ 20 comments factorio
- Deeplearning4j 0.9.0: Better ETL, Model Zoo, Training Optimizations https://deeplearning4j.org/releasenotes#zeroninezero 7 comments scala
- [R] Tune As You Scale: Hyperparameter Optimization For Compute Efficient Training https://arxiv.org/abs/2306.08055 3 comments machinelearning
- [R] Cerebras-GPT: Open Compute-Optimal Language Models Trained on the Cerebras Wafer-Scale Cluster https://arxiv.org/abs/2304.03208 39 comments machinelearning
- Calculate the optimal train configuration (python script) https://repl.it/repls/UncomfortableDarkkhakiProject 20 comments factorio
- "Internal optimism" Stafford will have new deal before start of Training Camp. http://www.nfl.com/news/story/0ap3000000820573/article/internal-optimism-stafford-will-have-new-deal-soon?campaign=twitter_atn 41 comments nfl
- A tip how to unload trains faster by optimizing inserter placement https://mods.factorio.com/mods/Bobingabout/bobinserters 4 comments factorio
- [P] DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer Models https://github.com/microsoft/DeepSpeed/tree/master/blogs/deepspeed-ulysses 5 comments machinelearning
- [R] Sophia: A Scalable Stochastic Second-order Optimizer for Language Model Pre-training https://arxiv.org/abs/2305.14342 6 comments machinelearning
- [Ian Rapoport] reports there is "some optimism" that Michael Thomas (ankle) will be cleared early in training camp. https://twitter.com/RapSheet/status/1549536227925303299?s=19 121 comments nfl
- We're running a challenge to optimize the scheduling of Swiss trains https://medium.com/crowdai/can-you-make-swiss-trains-even-more-punctual-ec9aa73d6e35 9 comments datascience
- [Bouhafsi] Still no Marco Verratti in PSG training in Lisbon. Still not looking good for Navas with his lesion. Mbappe trained normally, Big optimism for Silva https://twitter.com/mohamedbouhafsi/status/1294336363765145600?s=21 46 comments soccer
- MIT Researchers Propose a New Deep Reinforcement Learning Algorithm Trained to Optimize Doses of Propofol to Maintain Unconsciousness During General Anesthesia https://www.marktechpost.com/2022/02/16/mit-researchers-propose-a-new-deep-reinforcement-learning-algorithm-trained-to-optimize-doses-of-propofol-to-maintain-unconsciousness-during-general-anesthesia/ 4 comments reinforcementlearning
- [Sami Mokbel] There is positive news for Arsenal on the injury front amid optimism long-term absentee Jurrien Timber can return to full training next month https://www.dailymail.co.uk/sport/football/article-12993137/Declan-Rice-Arsenal-Nottingham-Forest-Crystal-Palace.html 9 comments soccer
- Optimizing Resistance Training Technique to Maximize Muscle Hypertrophy —flexibility is key, focus on Range of Motion (ROM) at long lengths. Eccentric/concentric tempos vary (2-8s). https://www.mdpi.com/2411-5142/9/1/9 73 comments science
- Ian Rapoport reports that DeVante Parker's new contract will not dampen any optimism of the Patriots landing DeAndre Hopkins, who's not expected to sign with a club until just before training camps kick off in late July. https://www.nfl.com/news/patriots-wr-devante-parker-agree-to-new-three-year-deal-worth-up-to-33m 15 comments nfl
- [R] Z-Code++: A Pre-trained Language Model Optimized for Abstractive Summarization - Microsoft Research 2022 - Outperforms the 600x larger PaLM-540B on XSum! https://arxiv.org/abs/2208.09770v1 4 comments machinelearning
- [Bob Nightengale] The Texas #Rangers show their optimism of the #MLB lockout ending before spring training, announcing that spring-training ticket packages will go on sale Monday. https://twitter.com/bnightengale/status/1469421198668935177?s=21 31 comments baseball
- [Heyman] Sources: MLB will make concessions on key economic issues, hoping to trigger more compromising by union. While owners are expressing optimism a deal will be struck to keep the season start date as March 31, the new offer still thought unlikely to avert a delay to spring training. https://twitter.com/JonHeyman/status/1491839384584339460?amp%3Bt=0s69LcYMHrOZ1hZUycvpTQ&s=20 118 comments baseball
- [Martino] Rob Manfred will speak publicly mid-day today. With news of this new proposal comes some cautious optimism that a spring training delay won't be extended. But then again I'm at the owners meetings -- let's see what the players have to say about the details when they become clear. https://twitter.com/martinonyc/status/1491791119897665549 34 comments baseball
- [R] Check the blog that introduces our recent #NeurIPS2021 work on nonconvex stochastic optimization! We propose stochastic Anderson mixing with theoretical guarantees and promising results in training neural networks. https://thumtblog.github.io/2021/10/30/stochastic-anderson-mixing/ 2 comments machinelearning
- The optimism around Kylian Mbappe starting against Manchester City is no longer true. Caution has remained, and tonight's training session at the Etihad is set to be decisive, with Mbappé aiming to take part in high intensity work. At present, he is uncertain to start. https://rmcsport.bfmtv.com/football/ligue-des-champions/manchester-city-psg-l-inquietude-grandit-pour-mbappe_av-202105030260.html 70 comments soccer
- @ElonMusk: Looks like a thin, flat, round UFO on a stick. Starlink Terminal has motors to self-adjust optimal angle to view sky. Instructions are simply: - Plug in socket - Point at sky These instructions work in either order. No training required. https://twitter.com/elonmusk/status/1214548764054216704 312 comments spacex
- [OC] A Comprehensive Engineering Analysis of the 3-Point Shot in Basketball: Player-Specific Optimization Kinematics Model and Methods for Improved Training and Performance Outcomes https://docdro.id/YXizUp9 15 comments nba
- Optimism can change brain volume, providing protection against emotional distress, study finds. The results support the hypothesis that personality traits and brain volume help protect against symptoms of anxiety - 'We can change the volume of the brain through experience and training.' https://www.inquisitr.com/5051313/study-optimism-can-change-brain-volume-providing-protection-against-emotional-distress/ 37 comments science