Shivam Kumar
PulseAugur coverage of Shivam Kumar — every cluster mentioning Shivam Kumar across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
Runway releases Praxis-1 open-weight robot action model
Runway has released Praxis-1, a 27-billion-parameter open-weight model designed for predicting future video and robot actions. This model achieved a 46.2% success rate on NVIDIA's RoboLab-120 leaderboard, outperforming …
-
Google DeepMind uses AI agents to formally verify Signal's cryptographic protocol
Google DeepMind has introduced AlphaProof Code, an AI-powered framework that formally verifies complex software. Researchers used this system to mathematically prove key security properties of Signal's cryptographic pro…
-
Inkling 2 Small: A Developer's Checklist Before Committing to a New Model
Thinking Machines Lab has released Inkling 2 Small, a 594B parameter model with 17B active parameters per token, scoring 31 on the Artificial Analysis Intelligence Index. Currently, access is limited to the Tinker platf…
-
GitHub Copilot Fusion automates AI model selection for developers
GitHub has introduced Copilot Fusion, a new feature designed to automate model selection for developers by routing tasks across different AI models based on specified effort levels. This approach aims to optimize cost a…
-
Apple M6 chips accelerate on-device AI, enabling local-first strategies
Apple has launched new MacBook Pro and iMac models featuring the M6 chip, significantly boosting on-device AI processing capabilities. These new machines offer substantial speed improvements for AI tasks, with the iMac …
-
DeepSeek V4.1-Flash slashes AI agent costs by 15x with smart routing
A developer significantly reduced AI agent inference costs by switching from GPT-4 class models to DeepSeek V4.1-Flash. By implementing a routing strategy that directs simpler tasks to the more cost-effective DeepSeek m…
-
Developer builds tiny 188M MoE LLM from scratch on free GPU
Shivam Kumar, founder of VisionQuantech, details his process for building a small, 188 million parameter Mixture-of-Experts (MoE) language model using only a free Nvidia T4 GPU. He employed a methodology called the Main…