In today's episode of programming horror… In the Python docs of random.seed() def, we're told
"If a is an int, it is used directly." [1] But if you seed with 3 or -3, you actually get the exact same rng object, producing the same streams. (TIL). In nanochat I was using the
CODE
-
Python Random Seed Bug: Identical Streams from Different Seeds
By
–
-
Gemini 3.0 and GPT-5 Codex Max for Full Stack Development
By
–
Pro Tip – Use Gemini 3.0 for visual understanding and front-end coding Combine with GPT-5 Codex Max for backend!
-
Claude AI assists with CUDA installation on Windows
By
–
OH MON DIEU !! SI C’EST PAS L’AGI, ÇA !! Claude vient de m’aider à fixer à 1 h du mat une installation de CUDA sur Windows, sur un support nightly pas encore adapté pour Blackwell !!!!!!!!!!!!!!!!!!!!!!!! Je suis carrément ému. La vie est trop belle. Vive la vie. Vive l’époque
-
Converting Qwen3 to Diffusion LLM: Block Diffusion Method
By
–
Open recipe to turn Qwen3 into a diffusion LLM 👀👀
— Maxime Labonne @ ICLR (@maximelabonne) 8 décembre 2025
> Swap the causal mask for bidirectional attention
> Source model matters a lot for performance
> Block diffusion (BD3LM) >> masked diffusion (MDLM)
> Light SFT with masking
Great work from @asapzzhou with his dLLM library! pic.twitter.com/ec2tSXAUA1Open recipe to turn Qwen3 into a diffusion LLM > Swap the causal mask for bidirectional attention
> Source model matters a lot for performance
> Block diffusion (BD3LM) >> masked diffusion (MDLM)
> Light SFT with masking Great work from @asapzzhou with his dLLM library! -
Pika API Now Available for Developer Integration
By
–
Get excited! Now you can add Pika magic to your products and applications. Introducing the Pika API hosted by @fal.
— Pika (@pika_labs) 8 décembre 2025
The partnership allows developers to seamlessly integrate Pika model 2.2, including its signature features Pikascenes and Pikaframes, through the turnkey Fal API… pic.twitter.com/8nUFGVHCetGet excited! Now you can add Pika magic to your products and applications. Introducing the Pika API hosted by @fal
. The partnership allows developers to seamlessly integrate Pika model 2.2, including its signature features Pikascenes and Pikaframes, through the turnkey Fal API -
Stanford AI Lab Postdoctoral Fellows Program Deadline December 15
By
–
Deadline approaching: Applications for the @StanfordAILab Postdoctoral Fellows Program end December 15! Whether you're working on robotics, RL, computer vision, genomics, or NLP – we want to hear from you. Join Stanford's fun and dynamic AI community: https://
bit.ly/43mRPEE -
Z AI Releases GLM-4.6V Models
By
–
Z AI released GLM-4.6V (106B) and GLM-4.6V-Flash (9B) open models! Both are available on Z AI Chat and APIs. Coding evals
-
AI21 Maestro: Orchestration Platform for Reliable AI Workflows
By
–
AI21 Maestro is @AI21Labs' orchestration platform for building reliable end-to-end AI workflows.
— AI21 Labs (@AI21Labs) 8 décembre 2025
It combines multi-step planning, automatic compute scaling, built-in validation, proprietary RAG, and execution graphs/scorecards to keep AI agents accurate and transparent in… pic.twitter.com/otJIYDOI4VAI21 Maestro is @AI21Labs
' orchestration platform for building reliable end-to-end AI workflows. It combines multi-step planning, automatic compute scaling, built-in validation, proprietary RAG, and execution graphs/scorecards to keep AI agents accurate and transparent in -
Gemini vs DeepSeek vs ChatGPT in UI Development
By
–
Here’s the simple takeaway: If you’re building front-end UIs with AI right now, Gemini is on another level. DeepSeek is close. ChatGPT is not even in the race. Crazy how the same “build a site” request exposes the strengths instantly: → ChatGPT thinks like a coder
→ -
New algorithm done by LLM to activate in a month
By
–
Hah, the new algorithm that turns on in a month or so is completely LLM-done here. So in a month you will be wrong.