Keep on to blur preview images; turn off to show them clearly
RT @XiaomiMiMo: ⚡ Faster than Fast. Designed for Agentic AI. Introducing Xiaomi MiMo-V2-Flash — our new open-source MoE model: 309B total p…
Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI builders
RT @MaziyarPanahi: holy shit! 💩 > 310B MoE > 150 tokens/s > Hybrid Sliding Window Attention > Trained on 27T tokens, FP8 mixed percision >…
Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI builders
Anyway, I'm excited for anyone who competes with me to go back to typing their code out by hand. I love that for me.
Author. Coder. CTO. θηριομάχης. Building: https://t.co/otXT4Wy6WR. Writing: https://t.co/dBPBtyCIHw.
RT @MannyBernabe: 🔴 We're live! Stripe live build session part 2! @raymmar_ is walking us through adding payment and subscription options…
ceo @replit. civilizationist
To the quality point, I will say that if you're getting a lot of AI slop in your code from working this way then you're doing it wrong. You're just not managing context very well &/or not paying close enough attention to output as the model types for you.
Anyway, I'm excited for anyone who competes with me to go back to typing their code out by hand. I love that for me.
I feel like I live in an alternate universe. I ship a lot of code, and it's 99% prompt => PRD/plans => code. I just don't type code anymore except very rarely. And no I'm not in the novelty phase bro I've been prompting coding models since it was even possible.
So I don't get this post or any of the replies that agree with it. And I see similar to this on the TL all the time. My TL has bifurcated into people who prompt all their code into existence, and people who don't for whatever quality or other reasons.