Your curated collection of saved posts and media
people forget how influential Palm webOS was. lots of modern iOS uses similar patterns. palm was basically the OpenAI of their time. such talent density that Steve Jobs had to threaten them to stop poaching. ⒠horizontal app card multitasking w/ flick up to quit ⒠edge swipe⦠https://t.co/xajSL950wX
I know more than one people who can guess betas and correlations visually. Once I met a person (not joking) who guessed the correlation using one data point. So I decided to train my eyes. You can do it too and influence people and win friends. https://t.co/U8rpIddlZE
https://t.co/D8HbMthQDN
me when a new model drops https://t.co/aa3rjjfxVu
Recently I've been telling @AmpCode to create PRs and it surprised me today by using the GitHub cli for the first time https://t.co/K7ophHbGAK
Youβre not falling behind. One Claude Code agent and typing is still just fine Ignore FUD https://t.co/CaZ1YGtDRQ
git work trees made easy with this bash script just run new-features <branch-name> --uv --bun and it automatically installs deps, and goes to the right directory :) https://t.co/oydAXkhisL
hermes 405b is a great bot https://t.co/ETYIZDNNnD
New Course: ACP: Agent Communication Protocol Learn to build agents that communicate and collaborate across different frameworks using ACP in this short course built with @IBMResearch's BeeAI, and taught by @sandi_besen, AI Research Engineer & Ecosystem Lead at IBM, and⦠https://t.co/ykhi6eHH8N
lol data efficient refactoring with amp <3 https://t.co/l3I6hH7jo2
Sped up import time by 700% after learning about getattr in python :) https://t.co/9sj4WNtaxa
things are getting weird https://t.co/NKA9CGOfbo
Grok 4 benchmarks https://t.co/9JGZIPrCE7
New Course: Post-training of LLMs Learn to post-train and customize an LLM in this short course, taught by @BanghuaZ, Assistant Professor at the University of Washington @UW, and co-founder of @NexusflowX. Training an LLM to follow instructions or answer questions has two key⦠https://t.co/ssqCqYqK4K
Should I use "ready-to-use" evaluation metrics? Please, for the love of God - DON'T DO IT (unless you really know what you're doing). https://t.co/8OILKKculg
Did you know that you can load checkpoints from Hugging Face Hub directly into a KerasHub model? That means you can load model weights from other frameworks, like using PyTorch weights with JAX or TensorFlow. Check out this blog post to learn how β https://t.co/1DCkxet4NB https://t.co/BKqhXgRBIU
To clarify, Matt is using the Claude Code provider in Cline, meaning that he can tap into his Claude Max (or Pro) subscription and not pay any additional costs per request. & in turn, run $0 tasks using Sonnet or Opus-4 our guide on using the Claude Code provider below β https://t.co/sW7UjwlV2s
MemAgent MemAgent-14B is trained on 32K-length documents with an 8K context window. Achieves >76% accuracy even at 3.5M tokens! That consistency is crazy! Here are my notes: https://t.co/G3LTUpGyhF
Update: @cluely filed a DMCA takedown for my tweet about their system prompt, alleging that it contained "proprietary source code" Making legal threats against security researchers is not a good look, and I encourage Cluely to reflect on this and open doors to researchers. 𧡠https://t.co/G0KSod0irx
Distributed training unlocks your ability to train your own SOTA models (as long as data is good and you have compute for long enough). We're seeing this today with the open release of SmolLM3: 3B parameters trained on 11 trillion tokens on 384 H100's for 24 days. Come learn⦠https://t.co/cjiRJ86paI
This is what the ideal grocery store looks like. Minimally processed (NOVA Group 1) food only (no "edible food-like substances"), organic, local, fresh. Food should not be more complex than this, yet I don't believe this exists. https://t.co/joHUslcdM3
Great advice from @BEBischof on "What Evals Framework should you use" πΆοΈ He's a guest speaker in our evals course! https://t.co/GZLIbA2kWq https://t.co/Kowjz1X7Ar
Excited to introduce Reka Vision, an agentic visual understanding and search platform. Transform your unstructured multimodal data into insights and actions. https://t.co/b1TECnjeCW
Context Engineering Guide is now part of the Prompt Engineering Guide. π₯ Nicer format. We've also been writing guides on other fire topics, such as deep research, reasoning LLMs, and image generation. I will be expanding the guide further in the coming days. Stay tuned! https://t.co/IZdVDa9icc
AI is already causing layoffs and slowing hiring, especially for entry level white collar jobs, in multiple industries. Similar to a sluggish post GFC job market, I think we will see a wave of new entrepreneurs who build the future they want to participate in. https://t.co/15QHH52CUH
I got permission to publish this massive EI Evals FAQ (PDF). It's like a bible for AI engineers and AI PMs. @HamelHusain and @sh_reya answer the most common questions they got while teaching 700+ students. And share 30+ free videos, posts, and resources: 𧡠https://t.co/rJrVi3Pr8x
How can we unlock generalized reasoning? β‘οΈIntroducing Energy-Based Transformers (EBTs), an approach that out-scales (feed-forward) transformers and unlocks generalized reasoning/thinking on any modality/problem without rewards. TLDR: - EBTs are the first model to outscale theβ¦ https://t.co/CHUVzp34hF
LLM built by geologist for the geologist #AI2S https://t.co/XbOkeWuyuW https://t.co/mKtN5Xm1KB
Build citable AI applications with @AnthropicAI's tool calling and citations feature in LlamaIndex π LlamaIndex now supports @AnthropicAI's server-side tool calling with automatic citations, making it easy to build AI apps that can cite their sources and provide traceableβ¦ https://t.co/fK0CcfEOc5
so Chrome 138+ onwards ships Gemini Nano for every user*, putting a local LLM in 3.7 billion monthly active users of Chrome wrote a personal guide today for building with Gemini Nano - mostly for me, might help you if you're not suuper strong at javascript or parsing google's...β¦ https://t.co/yhgog5Wcrk
Here is @orionweller 's talk posted on yt. BTW 4o one-shotted this thumbnail directly from the talk's transcript! (annotated notes coming soon) https://t.co/OTJaKvcfAc https://t.co/jkXCgzlbcS
Today we published a targeted transparency framework for frontier AI development. Our framework focuses on major frontier model developers while exempting startups and smaller developers to avoid burdening the broader ecosystem. https://t.co/UrU8Vjtsm4
