Joye Lin

programmer

[Kilian Lieret @KLieret]

Releasing mini, a radically simple SWE-agent: 100 lines of code, 0 special tools, and gets 65% on SWE-bench verified!
Made for benchmarking, fine-tuning, RL, or just for use from your terminal.
It’s open source, simple to hack, and compatible with any LM! Link in 🧵

[Jiayi Pan @jiayi_pirate]

We reproduced DeepSeek R1-Zero in the CountDown game, and it just works

Through RL, the 3B base LM develops self-verification and search abilities all on its own

You can experience the Ahah moment yourself for < $30
Code: https://t.co/UcGKN2SVGj

Here's what we learned 🧵

[Ivan Velichko @iximiuz]

docker run -p 8080:80 nginx:alpine

...is the same as

docker run -p 0.0.0.0:8080:80 nginx:alpine

The command opens port 8080 on all server interfaces, including the public ones (if any).

Why is it bad? If you close all ports by default using something like ufw, Docker's iptables rules will take precedence, and the port will become open regardless of your default firewall policy - something you likely don't want to happen.

Instead, most of the time, you should do the explicit:

docker run -p 127.0.0.1:8080:80 nginx:alpine

This is a good example of a poor choice for the default behavior. Was it motivated by the desire to increase the adoption during the early days of Docker, or was it just an oversight?

[Ivan Velichko @iximiuz]

Confused by Docker image options for Node.js? Dealing with huge containers and noisy vulnerability scans?

Been there, too! 🙈

Check out my tutorial on choosing the optimal Node.js base image for your app:

https://t.co/wdn4oAkl3G

[Alex Garcia @agarcia_me]

sqlite-vec v0.1.0 is out! Vector search for SQLite that runs everywhere!

- Supported on MacOS, Linux, Windows, browser WASM, Raspberry Pis
- Install with Python, Node.js, Ruby, Go, Rust, Datasette...
- Float and binary vectors supported

Try it out!

https://t.co/RQxrvuaTza

[AI at Meta @AIatMeta]

Today we’re announcing Meta LLM Compiler, a family of models built on Meta Code Llama with additional code optimization and compiler capabilities. These models can emulate the compiler, predict optimal passes for code size, and disassemble code. They can be fine-tuned for new optimizations and compiler tasks.

@HuggingFace repo ➡️ https://t.co/9URAr9sn5E
Research paper ➡️ https://t.co/nIYvWHqm1D

LLM Compiler achieves state-of-the-art results on code size optimization and disassembly. This work shows that AI is learning to optimize code and can assist compiler experts in identifying opportunities to apply optimizations.

We’re releasing LLM Compiler 7B & 13B models under a permissive license for both research and commercial use in the hopes of making it easier for developers and researchers alike to leverage this in their work and carry forward new research in this space.

[Guilherme Penedo @gui_penedo]

We are (finally) releasing the 🍷 FineWeb technical report!

In it, we detail and explain every processing decision we took, and we also introduce our newest dataset: 📚 FineWeb-Edu, a (web only) subset of FW filtered for high educational content.

Link: https://t.co/MRsc8Q5K9q

[Andrej Karpathy @karpathy]

Awesome and highly useful: FineWeb-Edu 📚👏
High quality LLM dataset filtering the original 15 trillion FineWeb tokens to 1.3 trillion of the highest (educational) quality, as judged by a Llama 3 70B. +A highly detailed paper.

Turns out that LLMs learn a lot better and faster from educational content as well. This is partly because the average Common Crawl article (internet pages) is not of very high value and distracts the training, packing in too much irrelevant information. The average webpage on the internet is so random and terrible it's not even clear how prior LLMs learn anything at all. You'd think it's random articles but it's not, it's weird data dumps, ad spam and SEO, terabytes of stock ticker updates, etc. And then there are diamonds mixed in there, the challenge is pick them out.

Pretraining datasets may also turn out to be quite useful for finetuning, because when you finetune a model into a specific domain (as is very common), you slowly lose general capability. The model starts to slowly forget things outside of the target domain. But this is not only restricted to knowledge; You also lose more general "thinking" skills that the original data demanded, but your new domain might not exercise. i.e. in addition to the broad knowledge fading, those computational circuits also slowly degrade. So there are likely creative ways to blend the pretraining and finetuning stages.

[WizardLM @WizardLM_AI]

🔥Today we are announcing WizardLM-2, our next generation state-of-the-art LLM.

New family includes three cutting-edge models: WizardLM-2 8x22B, 70B, and 7B - demonstrates highly competitive performance compared to leading proprietary LLMs.

📙Release Blog: https://t.co/W8Suq2vi6w

✅Model Weights: https://t.co/FWJs94FmB6

[Abhik Roychoudhury @AbhikRoychoudh1]

AutoCodeRover: Autonomous Software Engineer

Resolves 22% of Github issues in SWE-benchlite in &lt;10 mins at minimal LLM cost ~$0.5
Works on program representation of Abstract Syntax Tree, and exploits program structure in the form of classes/methods/APIs
https://t.co/o10aqonR2z

[Matt Shumer @mattshumer_]

Introducing `claude-researcher` 📈

A powerful Claude 3 research agent that delivers thorough reports in record time.

Just provide an topic, and a chain of AIs with **access to Google** will generate an incredibly comprehensive report for you.

And it's open-source!

[carlos @closji]

SWE-Agent is an open-source software engineering agent with a 12.3% resolve rate on SWE-Bench!

Check out SWE-agent in action at https://t.co/1NNL526gMy
Repo: https://t.co/LsgeVvD1UC

[the tiny corp @__tinygrad__]

Preorders will soon have a decision to make.

[Cybersecurity @ NIST @NISTcyber]

📣NEW BLOG | Travel Update! The @NIST #CybersecurityFramework (CSF) 2.0 is HERE…Along with Many Helpful Resources…

https://t.co/pHcS62W1A2

[Matt Shumer @mattshumer_]

Highest alpha secret in AI right now:

If you provide ~10 examples to Claude 3 Haiku…

it’ll often outperform Claude 3 Opus, and far outperform GPT-4

at a fraction of the cost, with blazing fast speeds

[Nous Research @NousResearch]

Today we are announcing the release of a new model, Genstruct 7B. Genstruct 7B is an instruction-generation model, designed to create valid instructions given a raw text corpus. This enables the creation of new, partially synthetic instruction finetuning datasets from any raw-text corpus.

Download the model + notebook on our HuggingFace page, here: https://t.co/HyVFDwCeLN

Inspired by the paper, Ada-Instruct, we took this approach further by grounding the generations in user-provided context passages. Further, the model is trained to generate questions involving complex scenarios that require detailed reasoning, allowing for models trained on the generated data to reason step-by-step.

This project was led by @Euclaise_ at Nous Research.

[Deno @deno_land]

happy lunar new year

[Deno @deno_land]

git log --since "2023-01-01" --until "2024-01-01"

https://t.co/BFxk68E3Kj

[Killer Shell @_killer_shell]

The complete CKS course is now FREE ON YOUTUBE https://t.co/HTMfPUqAUz.
We decided to open the material up to even more people through Youtube. Best luck with your CKS journey!

#cks #kubernetes #k8s #security #devops #devsecops

[Animo @AnimoSolutions]

The framework formerly known as Aries Framework JavaScript has moved to the @OpenWalletFdn and is in the process of becoming more modular and customizable than ever. To keep an eye on new developments, check out the GitHub https://t.co/c7aG0gsRqk