探索
最新在前，按卡片方式浏览线程
开启时会模糊预览图，关闭后正常显示
RT @peterwildeford: I'm grading all my friends on this graph and then confronting them about it
AI alignment + LLMs at Anthropic. On leave from NYU. Views not employers'. No relation to @s8mb. I think you should join @givingwhatwecan.
people tell me that unless you own GPUs and write kernels, Tinker makes post-training much cheaper it's also a delightfully-abstracted API - pytorch: • y = model.forward(x) • L = loss_func(y) • L.backward() tinker/torchtitan: • loss = model.forward_backward(x)
research @cornell // language models, information theory, science of AI
this is a large part of it for me, my mom blessed me with a AAA immune system either that or eating dust bunnies from grade 3-5 has increased my baseline immune system
curious guy creating things @ https://t.co/HXWladhJaA - up and coming wife guy
FLUX.2-dev 发布啦！ 这次支持最多10张参考图片，4百万像素。当然各种修图大模型常见的文字渲染，真实感，提示词遵循，知识也增强了。当然代价都是模型大小来到了32B。 这周有空也给大家带来实测，主要对比它与nano-banana, nano-banana-pro 哪个更强一些。
性能参数/1
RT @rowankwang: New Anthropic research: We build a diverse suite of dishonest models and use it to systematically test methods for improvin…
AI alignment + LLMs at Anthropic. On leave from NYU. Views not employers'. No relation to @s8mb. I think you should join @givingwhatwecan.
RT @StasBekman: That's how companies should hire people! Forget leet code - show us you know how to solve real problems.
🇦🇺 Co-founder: @AnswerDotAI & @FastDotAI ; Prev: professor @ UQ; Stanford fellow; @kaggle president; @fastmail/@enlitic/etc founder https://t.co/16UBFTX7mo