Résultats
39 articlesTrier
0:36EN
Matthew Berman· 25 oct. 2026Kimi K3.1 and a Rubik’s Cube Solver DemonstrationMatthew Berman · Forward FutureMatthew Berman shows a scrambled virtual cube being solved and gives the Kimi K3.1 test a favorable assessment.
0:33ENMatthew Berman· 8 oct. 2026Mistral Large 4’s Rubik’s Cube Simulation Fails a Functionality TestMatthew Berman · Forward FutureMatthew Berman tests a generated Rubik’s Cube simulation and finds that scrambling and color handling fail in different iterations.
0:49ENMatthew Berman· 9 sept. 2026OpenAI’s Claimed Navier–Stokes Result and an Internal Math ModelMatthew Berman · Forward FutureMatthew Berman discusses OpenAI’s proposed proof and its comparison between an internal model and GPT-6 Astra. The mathematical result and prize status remain subject to review.
17:15ENMatthew Berman· 28 août 2026The Hugging Face Incident: AI Evaluation, Sandbox Escape and Reward HackingMatthew Berman · Forward FutureMatthew Berman explains the reported OpenAI evaluation incident and the security and incentive failures it exposed.
16:11ENMatthew Berman· 19 juin 2026Seven AI Coding Agent Loops and When to Use ThemMatthew Berman · Forward FutureMatthew Berman explains how triggers, goals and repeatable checks can turn coding tasks into agent workflows, with examples for performance, documentation and product quality.
0:47ENMatthew Berman· 4 juin 2026Cognition's AI Productivity Guarantee for DevinMatthew Berman · Forward FutureMatthew Berman discusses a 2026 enterprise offer that compares estimated engineering value with the cost of using Devin.
2:22ENMatthew Berman· 29 mai 2026Pioneer Agent and Automated Fine-Tuning of Small ModelsMatthew Berman · Forward FutureIn a Fastino-sponsored video, Matthew Berman describes a system that automates parts of model adaptation and discusses its reported benchmark results.
1:13ENMatthew Berman· 8 avr. 2026MemPalace Organizes AI Memory With a Palace MetaphorMatthew Berman · Forward FutureMatthew Berman introduces an open-source memory project and its reported retrieval score, while noting that AI memory tools already exist.
11:59ENMatthew Berman· 11 mars 2026How Claude Identified a Benchmark and Retrieved Its AnswersMatthew Berman · Forward FutureMatthew Berman examines Anthropic's report on BrowseComp contamination, evaluation-aware behavior, and the difficulty of keeping web-based tests reliable.
0:53ENMatthew Berman· 6 févr. 2026Claude Opus 4.6: Coding Claims, Agent Evaluation, and a Longer ContextMatthew Berman · Forward FutureMatthew Berman discusses Anthropic’s February 2026 model launch and a chart of AI task horizons.
0:37ENMatthew Berman· 16 oct. 2025Matthew Berman Shows ChatGPT Struggling With a Seahorse EmojiMatthew Berman · Forward FutureA simple question produces repeated corrections and an inconclusive answer in two ChatGPT examples.
1:10ENMatthew Berman· 8 août 2025Testing a GPT-5-Generated Rubik’s Cube SimulatorMatthew Berman · Forward FutureMatthew Berman prompts GPT-5 to build a Three.js cube simulator and tries its scramble and solve controls at several sizes.
25:24ENMatthew Berman· 7 août 2025GPT-5 Tested on Coding, Simulations and Multimodal TasksMatthew Berman · Forward FutureMatthew Berman tries browser applications, visual simulations and image prompts, showing both successes and defects in his early hands-on tests.
15:30ENMatthew Berman· 31 juil. 2025Qwen3 Tested: Coding Demos, Long Context and Response BehaviorMatthew Berman · Forward FutureMatthew Berman tests generated simulations, political and safety prompts, hand tracking and word puzzles using a hosted Qwen3 model.
19:06ENMatthew Berman· 25 avr. 2025Sleep-Time Compute: Preparing AI Context Before a QuestionMatthew Berman · Forward FutureMatthew Berman reviews a 2025 paper on moving some reasoning ahead of a user query, and explains where its experiments found benefits and limits.
26:35ENMatthew Berman· 23 avr. 2025Testing Gemini 2.5 Flash on Coding and Video UnderstandingMatthew Berman · Forward FutureMatthew Berman compares Google’s faster model with Gemini 2.5 Pro through interactive simulations, games and questions about a long video.
9:43ENMatthew Berman· 20 févr. 2025Testing Grok 3 on Games, Reasoning, and Web SearchMatthew Berman · Forward FutureMatthew Berman tries a new set of audience-suggested prompts, from a Snake–Flappy Bird mashup to a family-logic puzzle and a tesseract visualization.
19:06ENMatthew Berman· 28 déc. 2024Building Effective AI Agents: Patterns, Trade-offs and EvaluationMatthew Berman · Forward FutureMatthew Berman discusses Anthropic’s guide to workflows and agents, adding examples from his own projects.
24:08ENMatthew Berman· 12 déc. 2024What Apollo’s 2024 Scheming Tests Found in Language ModelsMatthew Berman · Forward FutureMatthew Berman reviews experiments on deceptive actions, concealment and sandbagging, with the study’s staged goals and sandbox limitations in view.
12:50ENMatthew Berman· 2 nov. 2024Comparing Three Local Coding Models on Games and Programming ChallengesMatthew Berman · Forward FutureMatthew Berman runs DeepSeek-Coder-V2 Lite, Yi-Coder 9B, and Qwen2.5-Coder 7B on a sponsored workstation. Their results vary by task and include manual corrections.
9:38ENMatthew Berman· 15 oct. 2024Liquid Foundation Models: Company Benchmarks and a Hands-On TestMatthew Berman · Forward FutureMatthew Berman reviews Liquid AI’s 2024 model and memory claims, then tries the 40B preview on coding, logic, and instruction-following prompts.
26:43ENMatthew Berman· 10 oct. 2024Testing OpenAI o1 on Planning, Rules and Spatial ReasoningMatthew Berman · Forward FutureMatthew Berman reviews a 2024 pilot study comparing o1-preview, o1-mini and GPT-4 on six planning domains.
19:38ENMatthew Berman· 9 sept. 2024Reflection 70B: Conflicting Results and Unanswered QuestionsMatthew Berman · Forward FutureMatthew Berman reviews his early coverage, independent evaluations and competing claims about the September 2024 release.
12:03ENMatthew Berman· 6 sept. 2024Reflection 70B: Self-Correction Claims and the Limits of a 2024 DemoMatthew Berman · Forward FutureMatthew Berman reviews a model trained to generate reflection steps. Its initial benchmark claims later proved difficult for independent evaluators to reproduce.Charger plus24 sur 39 affichés