Hugging Face is where much of the machine learning community publishes and finds its datasets, while DuckDB is the in-process analytical database that queries files like CSV and Parquet directly, with no server or warehouse to install or run. 🤗 🦆
Did you know that, since #DuckDB v0.10.3 (~May 2024), you can point a SELECT at a dataset on the Hugging Face Hub, using the DuckDB `hf://` protocol, and query it, without downloading it first?
This blog post covers how that integration works and the use cases it fits: duckdb.org/2026/09/25/huggin…
8
15
114
5,466
p(doom)scroll. TikTok, but every video is a robot learning to do chores! 🤖
@huggingface shipped @huggingface/lerobot, a tiny JS library that streams any robot dataset on the Hub straight into the browser.
1
5
302
Point it at any LeRobot dataset and get back per-episode video clips, task captions, and joint data. Works in the browser and Node, no downloads, just ranged requests.
npmjs.com/package/@huggingfa…
1
82
.@jaredpalmer is a must-follow on @huggingface. doing some epic work out in the open!
> 5 open decision models of various sizes
> open dataset used for training
> finetuning scripts + agent skill
> browser demo
UPDATE: Kev-0.6B, 4B, and 8B are now available. Kev is a family of small open source Jev-like decision models you can train and run yourself.
This new family is based on Qwen3 using the same LoRA + small pointer head technique as before, but scaled up.
Out of domain, on data Kev never trained on: Kev-8B 79.6%, Jev 85.7%.
• Drop-in TypeSafe System One API; their SDK works with one `base_url` change
• Kev-4B serves on a 32 GB Mac in bf16: ~300 ms for five questions, ~40 ms on an H100
• Repeated documents hit a KV cache: 2-2.5x faster
• Apache 2.0 License. Kev-4B trains in 40 minutes on one H100. Kev-8B in 83 minutes.
Code, weights, evals: github.com/jaredpalmer/kev
3
1
12
789
Agent Traces on the @huggingface hub now come with a receipt 🧾
Every trace shows tokens, cache hit rate, and cost for every run!
5
9
43
26,615
huggingface.co/docs/hub/en/a…
it works out of the box with your jsonl files in buckets and datasets!
5
373
> “…it writes real regression tests instead of mock-based ones”
Wow! It’s Christmas 🎉
In our evaluations, we see that GPT-6 Sol reads about 17% less context than GPT-5.6 Sol to reach the same score, and rarely leaves a repo with failing tests after its last edit. GPT-6 Luna’s biggest gains are at low and medium reasoning effort, where it writes real regression tests instead of mock-based ones.
Read more: devin.ai/blog/gpt-6-sol-luna
1
4
425
GPT-6-Sol and Luna priced 2x cheaper than the 5.6 series!
So many great use cases for luna at that price. OpenAI keeps pushing the envelope on price!
1
9
488
Caleb retweeted
total MCP victory, some quick misc thoughts about why MCP is so much better than CLIs:
- indexable tool catalog letting agents scale to unlimited tools
- no requirements to be running a full sandbox
- consistent auth across all MCPs rather than each CLI inventing its own auth
- multi account support for all MCPs unlike CLIs
- implementations like code mode let the model know what will be returned allowing for super efficient token usage unlike CLIs
the reasons it took this long for MCPs to finally have their moment is mostly due to bad MCP implementations in clients:
- you had to restart your whole client to use an mcp (no hot reloading)
- agents weren't as familiar with debugging mcps as they were CLIs, so it was a lot easier for people to get set up using them
but over the past year, things like codex and claude plugins have all been using MCP under the hood, i.e computer use is an MCP, i believe claude artifacts are an MCP app, just the silent steady adoption
there is still an element of MCPs that is 'this MCP could've been an OpenAPI spec' but that'll go away as things like triggers get more adoption
at the end of the day, what's important to realize is while yes there are these difference between CLIs / MCPs / etc they're all just different ways of doing tool calling, and you can do some combination of lazy loading, searchable tools, and filtering to build efficient harnesses
I was not expecting things to go this way, but I think MCPs are better than CLIs for most integrations.
The models have gotten much better at tool calling, we can defer tools & MCP is now stateless.
If you need to compose/filter data, add params like query to your MCP tools.
95
59
1,016
183,991
Caleb retweeted
Introducing: a coding agent (Pi) running entirely in your browser using a 2B model on WebGPU 🤯
MiniCPM5-2B + Pi, powered by Transformers.js + WebGPU + 4-bit ONNX weights. All previous attempt to create this failed but MiniCPM5 seems to make it usable.
Available now on Hugging Face 👇
71
144
1,250
89,381
Does everyone just pin their favorite provider in OpenRouter when using for agentic tasks?
With long sessions, cache misses can be pretty expensive. Not to mention input cache pricing can vary widely by like 5x for different providers.
2
2
570
The open model inference space still has so much room to improve still. The default user experience is not very good:
> for popular models, most providers are rate limited, unavailable, or have super poor TPS
> prompt cache pricing can be really not competitive to closed weights sometimes
> prompt caching can very greatly by provider (a couple don't even support it lmao)
1
132
Caleb retweeted
I can’t stress enough how little an idea matters compared to the agency of the people executing the idea.
I have had the privilege of knowing and sometimes even working with some of the most successful people (by various metrics).
The difference between mediocre and excellent work and outcomes is predominantly one of agency.
In practice this means: they dont wait for things to happen to them they go out and make things happen for them.
They don’t wait for someone else to do something, for someone to teach them, for someone to give them the path, etc. They just go out and find a way to do it.
I think the single biggest superpower these people have is the realization/belief that the world around them is completely mutable. Most everything that happens is because a person made it happen.
I used to tell people to look around the room you’re sitting in. Look at everything. Every noun. It almost all exists because a person willed it into existence. Nothing is stopping you from doing the same.
I see people online all the time dismissing someone else’s success because “I had that idea first” or whatever. I mean… yeah? If so then the difference is… you. So a bit of a self own whenever I hear that.
Number one tip: act with agency.
228
1,279
11,857
682,076
This looks incredibly cool. Need this in the Codex app!
Introducing Projects, a new way of working in Cursor.
Rather than creating a chat for every task, you work with a coordinator agent in a single, persistent thread.
Like @bot, your agent is always on, proactively manages work with subagents, and improves over time.
3
11
1,359
self hosting maps is a killer use case for @huggingface buckets!
🌍 entire planet in one 137.8 GB PMTiles file
🪣 S3 compatible using @bunjavascript’s S3 client
🔎 only fetches the pieces you need as you pan
super cool that this just works!!
6
4
28
6,274
HF Buckets are S3 compatible so using @bunjavascript’s S3 client makes this incredibly easy!!
huggingface.co/docs/hub/stor…
1
3
225
Protomaps + PMTiles make this super simple to self host. PMTiles figures out which byte ranges to fetch for the tiles you’re viewing
docs.protomaps.com/
1
154