Jim Deola retweeted
Hi, I'm Felix. I'm an eng lead working on Claude. I love watching people take flight and do things they never thought possible.
We often describe that as "being more ambitious" when using AI, which is tricky advice to follow. I wrote a tutorial:
felixrieseberg.com/how-to-be…
Jim Deola retweeted
Today, we are introducing Inkling.
Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.
thinkingmachines.ai/news/int…
Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵
Jim Deola retweeted
The first experimental evidence of recursive self-improvement (RSI).
Autoresearching the autoresearch agent for eight days.
The result beats the harness we hand-tuned for two years, on held-out benchmarks: 🧵(1/7)
Jim Deola retweeted
Exceptional people are rare. When you find someone wonderful, invest in them.
-When you find a great employee, pay them well.
-When you find a great friend, prioritize the relationship.
-When you find a great spouse, out-love them each day.
Relationships are probably the most important part of life. Take care of the great ones.
Jim Deola retweeted
It's crazy to me how much of software development is 'finding the right way to describe something that doesn't exist yet'.
It's a language problem. Figure out the right nouns and verbs, and you can build a system.
And LLM's are GREAT at helping you with this.
Jim Deola retweeted
Anthropic is donating the Model Context Protocol to the Agentic AI Foundation, a directed fund under the Linux Foundation.
In one year, MCP has become a foundational protocol for agentic AI. Joining AAIF ensures MCP remains open and community-driven. anthropic.com/news/donating-…
Jim Deola retweeted
i should NEVER be sharing this but whatever...
i'm dropping Machines Lab, featuring:
- 7-day prompt engineering program
- 20+ courses for n8n, AI SEO, Twitter, LinkedIn...
- 300+ super-prompts for content creation, copywriting, automation...
- AI Mentor trained on ALL the courses content
it's FREE because i want to build a closed circle of AI Machines, that's why i'm keeping access limited
reply "MACHINES" + RT if you want in (must be following me so i can DM)
Jim Deola retweeted
Here's how to build an MCP server with a single TypeScript file, and connect it to Claude Code.
No build step. No bullshit. Totally transformative.
Jim Deola retweeted
Replying to @traversymedia
Educators who don't update their courses do more damage than library authors who update their libraries.
Jim Deola retweeted
by-hand.ai/dpseek
Before the holidays, Leandro von Werra at HuggingFace suggested me to study multihead latent attention. He said that DeepSeek uses it and they built some of the best/cheapest models. That's what initially motivated me to study DeepSeek and prepare a lecture about it.
Since then, DeepSeek has been making waves, including the reasoning model they just released yesterday. The timing is great to do a lecture about it.
Q: Who is this lecture for?
A: Originally for my students but now for anyone with similar technical background and interest.
Q: What is my focus?
A: How it works, rather than what it can do. Many others have already commented on what it can do (i.e., benchmark results). I like to take you inside the blackbox to understand how it works instead.
Q: When?
A: Next Tuesday at 11am my local time (Mountain Time)
Q: How to join?
A: Use the registration link by-hand.ai/dpseek
---
How can you support AI by Hand ✍️?
1. Check out our store: store.byhand.ai
2. Repost ♻️
Jim Deola retweeted
If you keep showing up, you'll almost certainly break through — but probably not in the way you expected or intended. You need enough persistence to keep working and enough flexibility to enjoy success when it comes in a different form than you imagined.
Jim Deola retweeted
3 minute masterclass on the only "free lunch" in finance.
Taught at MIT by Jake Xia,
With no student debt needed.
Jim Deola retweeted
Behind The $215 Billion Data Center Boom: Here's Who Makes All The Key Components
zerohedge.com/ai/data-center…
Jim Deola retweeted
OpenAI Strawberry (o1) is out! We are finally seeing the paradigm of inference-time scaling popularized and deployed in production. As Sutton said in the Bitter Lesson, there're only 2 techniques that scale indefinitely with compute: learning & search. It's time to shift focus to the latter.
1. You don't need a huge model to perform reasoning. Lots of parameters are dedicated to memorizing facts, in order to perform well in benchmarks like trivia QA. It is possible to factor out reasoning from knowledge, i.e. a small "reasoning core" that knows how to call tools like browser and code verifier. Pre-training compute may be decreased.
2. A huge amount of compute is shifted to serving inference instead of pre/post-training. LLMs are text-based simulators. By rolling out many possible strategies and scenarios in the simulator, the model will eventually converge to good solutions. The process is a well-studied problem like AlphaGo's monte carlo tree search (MCTS).
3. OpenAI must have figured out the inference scaling law a long time ago, which academia is just recently discovering. Two papers came out on Arxiv a week apart last month:
- Large Language Monkeys: Scaling Inference Compute with Repeated Sampling. Brown et al. finds that DeepSeek-Coder increases from 15.9% with one sample to 56% with 250 samples on SWE-Bench, beating Sonnet-3.5.
- Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters. Snell et al. finds that PaLM 2-S beats a 14x larger model on MATH with test-time search.
4. Productionizing o1 is much harder than nailing the academic benchmarks. For reasoning problems in the wild, how to decide when to stop searching? What's the reward function? Success criterion? When to call tools like code interpreter in the loop? How to factor in the compute cost of those CPU processes? Their research post didn't share much.
5. Strawberry easily becomes a data flywheel. If the answer is correct, the entire search trace becomes a mini dataset of training examples, which contain both positive and negative rewards.
This in turn improves the reasoning core for future versions of GPT, similar to how AlphaGo’s value network — used to evaluate quality of each board position — improves as MCTS generates more and more refined training data.
Jim Deola retweeted
📊AI-Driven Research Assistant
This is an advanced AI-powered research assistant system that utilizes multiple specialized agents to assist in tasks such as data analysis, visualization, and report generation
Fully open source, uses LangGraph!
github.com/starpig1129/ai-da…

















