Résultats
1 769 articlesTrier
12:07EN
Matthew Berman· 2 févr. 2025TinyZero: A $30 Reinforcement-Learning Experiment on a Small Arithmetic TaskMatthew Berman · Forward FutureMatthew Berman explains how a pretrained Qwen model learned to search and check answers in Countdown, and why the experiment does not recreate DeepSeek-R1.
5:06:18EN
Lex Fridman· 2 févr. 2025PodcastDeepSeek's $5 million training claim and the five-year bet on computeAnime : Lex FridmanDylan Patel and Nathan Lambert walk Lex Fridman through DeepSeek-V3 and R1, export controls, TSMC, megaclusters, and who actually makes money in the AI race.
9:58ENMatthew Berman· 1 févr. 2025Matthew Berman Tests o3-mini on Games, Puzzles and Web SearchMatthew Berman · Forward FutureA 2025 hands-on test finds working game prototypes and several successful answers alongside a delayed Tetris row, an ambiguous puzzle and a search configuration mistake.
6:44ENMatthew Berman· 1 févr. 2025Testing DeepSeek Janus-Pro for Image Understanding and GenerationMatthew Berman · Forward FutureMatthew Berman tries the 7B multimodal model on diagrams, memes and image prompts using a sponsored cloud GPU demonstration.
9:58ENMatthew Berman· 31 janv. 2025OpenAI o3-mini Launch: Reasoning Effort, Benchmarks and a Snake TestMatthew Berman · Forward FutureIn January 2025, Matthew Berman reviews OpenAI’s reported results and tries two coding examples with different reasoning settings.
15:22EN
AI Explained· 31 janv. 2025o3-mini’s Early Results and the Rhetoric of an AI RaceA launch-day look at benchmarks, safety evaluations, and the pressures of competition.
57:11EN
Unsupervised Learning: With Jacob Effron· 30 janv. 2025PodcastA Million Tokens Cannot Write a Good JokeAnime : Jacob EffronPrinceton's Arvind Narayanan on why agent evals mislead, inference scaling hits a wall, and AI looks more like the internet than an overnight revolution.
1:50ENMatthew Berman· 30 janv. 2025Five AI Trends Discussed by Coveo’s Sébastien PaquetMatthew Berman · Forward FutureMatthew Berman summarizes a 2025 interview on production systems, smaller models, agents and trust.
0:52ENMatthew Berman· 29 janv. 2025Genie 2 Generates Interactive Worlds for AI ResearchMatthew Berman · Forward FutureMatthew Berman explains why Google DeepMind’s image-conditioned world model could help researchers create environments for agents, while the early demonstrations have clear limits.
56:07EN
Cognitive Revolution "How AI Changes Everything"· 29 janv. 2025PodcastBox Holds 100 Billion Files. Its CEO Says Enterprises Only Move for a 10xAnime : Nathan LabenzAaron Levie tells Nathan Labenz why AI excitement in the enterprise already outruns the early cloud — and why 98% reliability, seat-based SaaS, and incremental savings still will not ship.
5:27ENMatthew Berman· 29 janv. 2025Ways to Try DeepSeek R1: Hosted and Local Distilled ModelsMatthew Berman · Forward FutureMatthew Berman demonstrates DeepSeek chat, Groq’s hosted distill and a local model in LM Studio.
20:34ENMatthew Berman· 28 janv. 2025DeepSeek R1 and the Debate Over AI Costs and ComputeMatthew Berman · Forward FutureMatthew Berman examines 2025 reactions to DeepSeek, from training efficiency and market losses to open weights, privacy and the possibility of a demand rebound.
16:45ENMatthew Berman· 27 janv. 2025DeepSeek-R1 and the Debate Over AI Costs and ComputeMatthew Berman · Forward FutureMatthew Berman examines the January 2025 reaction to an open-weight reasoning model, disputed cost claims and competing views of future demand for AI chips.
14:55ENMatthew Berman· 25 janv. 2025How the Industry Reacted to OpenAI Operator in 2025Matthew Berman · Forward FutureMatthew Berman reviews early views on browser agents, their limitations, competing projects and examples of delegated tasks.
1:48:20ENCognitive Revolution "How AI Changes Everything"· 25 janv. 2025PodcastFrom 15% to 70% on Right-or-Wrong Scores: DeepSeek-R1 Laid Reinforcement Learning's Cards on the TableAnime : Nathan LabenzOn Inauguration Day 2025, DeepSeek and Moonshot dropped reasoning-model results that Nathan Labenz reads as proof the reinforcement-learning recipe is simple, public, and still climbing.
0:55ENMatthew Berman· 23 janv. 2025A Quick RTX 5090 Local AI TestMatthew Berman · Forward FutureMatthew Berman runs an 8B DeepSeek R1 distillation locally and reports the speed of one generation.
1:40:22ENCognitive Revolution "How AI Changes Everything"· 22 janv. 2025PodcastTwenty-atom crystals taught an AI the physics of a heartbeatAnime : Nathan LabenzOrbital Materials' Jonathan Godwin and Tim Duignan tell Nathan Labenz how a neural network trained on tiny crystals simulated a potassium ion channel on a single 2017 GPU.
15:10ENMatthew Berman· 22 janv. 2025DeepSeek R1 Tested on Coding, Reasoning and Sensitive QuestionsMatthew Berman · Forward FutureMatthew Berman runs the full R1 model on cloud GPUs and tries game programming, reasoning puzzles and politically sensitive prompts.
7:57ENMatthew Berman· 22 janv. 2025Stargate’s 2025 Announcement and the OpenAI–Microsoft PartnershipMatthew Berman · Forward FutureMatthew Berman reviews the planned US AI infrastructure project and a contemporaneous change to compute arrangements.
8:52ENMatthew Berman· 21 janv. 2025Industry Reactions to DeepSeek R1’s 2025 ReleaseMatthew Berman · Forward FutureMatthew Berman reviews early reactions to DeepSeek R1, from open-weight reasoning models and local hardware to benchmark results and API pricing.
1:24:57ENUnsupervised Learning: With Jacob Effron· 21 janv. 2025PodcastA 50,000-GPU cap and a 1,700-chip loopholeAnime : Jacob EffronSemiAnalysis founder Dylan Patel walks through the AI diffusion rule, China’s compute squeeze, and how xAI stood up a 100,000-GPU cluster in a closed Memphis appliance factory.
12:15ENMatthew Berman· 20 janv. 2025DeepSeek-R1: Open-Weight Reasoning, Benchmark Results and a First TestMatthew Berman · Forward FutureMatthew Berman examines DeepSeek’s January 2025 release, compares reported benchmarks and prices, and tries a handful of reasoning prompts.
2:09:55ENCognitive Revolution "How AI Changes Everything"· 18 janv. 2025PodcastThe Attacker Won All 70 Rounds Against Mind-Reading DefensesAnime : Nathan LabenzLuke Bailey, Erik Jenner, and Scott Emmons walk through a 50-page paper showing that probes, out-of-distribution detectors, and sparse autoencoders can be bypassed while the model still does the harm.
1:33:01ENCognitive Revolution "How AI Changes Everything"· 15 janv. 2025PodcastRare Diseases Are Common: How a Harvard Gene Hunter Uses o1-pro on the Cases Sequencing Already MissedAnime : Nathan LabenzDr. Catherine Brownstein of Boston Children's Hospital explains why cheap genomes made her job harder, why data sharing still beats any model, and where ChatGPT Pro actually gives her hours back.Charger plus24 sur 1 769 affichés