Sam Charrington
19K posts

Sam Charrington
@samcharrington
Machine learning & AI podcaster, community builder and all around enthusiast. Creator of the @TWIMLAI Podcast, TWIMLcon, TWIMLfest & the TWIML Solutions Guide.








After some initial attempts at using GPT-5.3-Codex via Cursor I thought the takes here suggesting how much better it was were highly exaggerated. But after spening some time working with it in the new Codex app, I'm impressed. The end-to-end experience feels much more accurate (i.e. it does what I want) and productive. Whether this reflects model or harness advancements we won't really know (the answer is likely both), but on vibes alone it really does feel like a step forward.


Some info about requests being routed from GPT-5.3-Codex to GPT-5.2 This is part of our effort to reduce cyber abuse risk: When our systems detect elevated cyber misuse risk, requests may be routed from GPT-5.3-Codex to GPT-5.2. Currently there's no UI in Codex to tell users when we're routing to 5.2—we'll add this! In some cases legitimate work may be incorrectly flagged — we’re actively tuning these systems and adding clearer notifications. In the meantime, if you’re doing defensive research or think you were misclassified, you can apply to regain access at chatgpt.com/cyber, or report refusals via /feedback.





Today, we're joined by @rdn_nikita, co-founder and CEO of @FlexionRobotics to discuss the gap between current robotic capabilities and what’s required to deploy fully autonomous robots in the real world. Nikita explains how reinforcement learning and simulation have driven rapid progress in robot locomotion—and why locomotion is still far from “solved.” We dig into the sim2real gap, and how adding visual inputs introduces noise and significantly complicates sim-to-real transfer. We also explore the debate between end-to-end models and modular approaches, and why separating locomotion, planning, and semantics remains a pragmatic approach today. Nikita also introduces the concept of "real-to-sim", which uses real-world data to refine simulation parameters for higher fidelity training, discusses how reinforcement learning, imitation learning, and teleoperation data are combined to train robust policies for both quadruped and humanoid robots, and introduces Flexion's hierarchical approach that utilizes pre-trained Vision-Language Models (VLMs) for high-level task orchestration with Vision-Language-Action (VLA) models and low-level whole-body trackers. Finally, Nikita shares the behind-the-scenes in humanoid robot demos, his take on reinforcement learning in simulation versus the real world, the nuances of reward tuning, and offers practical advice for researchers and practitioners looking to get started in robotics today. 🗒️ For the full list of resources for this episode, visit the show notes page: twimlai.com/go/760. 📖 CHAPTERS =============================== 00:00 - Introduction 04:07 - Is robot locomotion solved? 06:04 - Sim-to-real gap 08:58 - Adding semantics to policies 09:42 - Modular vs end-to-end architectures 10:29 - Planner model 12:21 - Adapting RL techniques from quadrupeds to humanoids 15:39 - Behind robot demos 18:09 - Humanoid robots in home environments 22:03 - Training approach 23:56 - VLA models 27:59 - Closing the sim-to-real gap 32:55 - Task orchestration using VLMs 36:38 - Tool use 38:10 - Model hierarchy 43:37 - Simulator versus simulation environment 44:57 - Combining imitation learning and reinforcement learning 46:42 - RL in real world versus RL in simulation 52:58 - Reward tuning and value functions in robotics 56:38 - Predictions 1:00:10 - Humanoids, quadropeds, and wheeled platforms 1:02:45 - Advice, recommended robot kits, and community pla





