alt Hacker NewsShow HN: Needle: We Distilled Gemini Tool Calling into a 26M Model
Hey HN, Henry here from Cactus. We open-sourced Needle, a 26M parameter function-calling (tool use) model. It runs at 6000 tok/s prefill and 1200 tok/s decode on consumer devices.
We were always frustrated by the little effort made towards building agentic models that run on budget phones, so we conducted investigations that led to an observation: agentic experiences are built upon tool calling, and massive models are overkill for it. Tool calling is fundamentally retrieval-and-assembly (match query to tool name, extract argument values, emit JSON), not reasoning. Cross-attention is the right primitive for this, and FFN parameters are wasted at this scale.
Simple Attention Networks: the entire model is just attention and gating, no MLPs anywhere. Needle is an experimental run for single-shot function calling for consumer devices (phones, watches, glasses...).
Training: - Pretrained on 200B tokens across 16 TPU v6e (27 hours) - Post-trained on 2B tokens of synthesized function-calling data (45 minutes) - Dataset synthesized via Gemini with 15 tool categories (timers, messaging, navigation, smart home, etc.)
You can test it right now and finetune on your Mac/PC: https://github.com/cactus-compute/needle
The full writeup on the architecture is here: https://github.com/cactus-compute/needle/blob/main/docs/simp...
We found that the "no FFN" finding generalizes beyond function calling to any task where the model has access to external structured knowledge (RAG, tool use, retrieval-augmented generation). The model doesn't need to memorize facts in FFN weights if the facts are provided in the input. Experimental results to published.
While it beats FunctionGemma-270M, Qwen-0.6B, Granite-350M, LFM2.5-350M on single-shot function calling, those models have more scope/capacity and excel in conversational settings. We encourage you to test on your own tools via the playground and finetune accordingly.
This is part of our broader work on Cactus (https://github.com/cactus-compute/cactus), an inference engine built from scratch for mobile, wearables and custom hardware. We wrote about Cactus here previously: https://news.ycombinator.com/item?id=44524544
Everything is MIT licensed. Weights: https://huggingface.co/Cactus-Compute/needle GitHub: https://github.com/cactus-compute/needle
Comments
Suggestion: publish a live demo of the "needle playground". It's small enough that it should be pretty cheap to run this on a little VPS somewhere!
I find this stuff super fascinating and been thinking about it myself. Maybe one could bootstrap tiny models on a rather 'pure' procedural data set. Neglecting [0] of course...
[0]: http://www.incompleteideas.net/IncIdeas/BitterLesson.html
Sounds interesting.
Got a bunch of errors trying to run it on CPU though. Very likely connected to me running this in a container (unpriv LXC), but figured for 26M CPU would suffice.
That M versus B is way too subtle. 0.026B is my suggestion
Looks like you need to open up access to https://huggingface.co/Cactus-Compute/datasets/needle-tokeni... - I get this error when trying to run the steps in your README:
> Repository Not Found for url: http s://huggingface.co/api/datasets/Cactus-Compute/needle-tokenizer/revision/main.
Hmm.. this might make it feasible to build something like a command line program where you can optionally just specify the arguments in natural language. Although I know people will object to including an extra 14 MB and the computation for "parsing" and it could be pretty bad if everyone started doing that.
But it's really interesting to me that that may be possible now. You can include a fine-tuned model that understands how to use your program.
E.g. `> toolcli what can you do` runs `toolcli --help summary`, `toolcli add tom to teamfutz group` = `toolcli --gadd teamfutz tom`
This is really cool. Any plans to release the dataset?
Can this be a Siri-like core? Set me a timer, tell me what’s the weather, etc. Here is transcribed text and available list of tools for the model to call, and voice the output.
FYI, distilling Gemini is explicitly against the ToS:
"You may not use the Services to develop models that compete with the Services (e.g., Gemini API or Google AI Studio). You also may not attempt to reverse engineer, extract or replicate any component of the Services, including the underlying data or models (e.g., parameter weights)."
This is very cool I'm going to try to carve out some time to try building this into my MOO system ( https://codeberg.org/timbran/moor / https://timbran.org/moor.html ) as alternative command parser front end.
[flagged]
[flagged]
This tool in the training data and playground surprised me:
Is there a reason to structure it like this instead of having it specify the time in a computer understandable way?Is the idea that there's be a harness which then calls the (or a different) model to convert the time to a structured duration or endpoint?