sparrow-micro-general (v1.1)

A 7.6 MB decision model with the full potion-base-8M vocabulary. It answers yes/no (noul), pick-one (choice) and rate-on-a-scale (score) questions about a text state, with calibrated probabilities, in 0.22 ms per decision on one laptop core with the C runtime.

Part of Sparrow Decider, a family of decision models from 47 MB down to a few KB. Same question and answer shapes as Strands Decider.

Unofficial community project. Not an AWS or Strands product.

Model ID	sparrow-micro-general-v1.1
Tier	micro (static embeddings + MLP)
File	sparrow-micro-general-v1.1.sparrow (7.6 MB, Python) · sparrow-micro-general-v1.1.spb (8.88 MB packed SPB1 blob, C)
Input	text
Runs on	Raspberry Pi, laptop
Dependencies	numpy (Python) · C99 + libm, no malloc (C)

Intended use

• Short text decisions (routing, gating, urgency) when you have a few MB to spare and want the full vocabulary instead of nano's pruned one; sub-millisecond on a CPU.
• Its MASSIVE domain routing (0.635) is higher than the nanos' (0.50-0.518).

Out of scope: open-ended language far from its training data, yes/no judgement that needs reasoning, and being the only safety mechanism on a physical system.

How to use

Python

pip install sparrow-decider huggingface_hub

from huggingface_hub import hf_hub_download
from sparrow_decider import Decider

path = hf_hub_download("sparrow-decider/sparrow-micro-general", "sparrow-micro-general-v1.1.sparrow", revision="v1.1")
d = Decider.load(path)
d.ask("User: please move my 3pm meeting with Sam to Friday", {
    "tool": {"type": "choice", "instructions": "Which tool should the agent call?",
             "criteria": {"weather": "get a weather forecast", "calendar": "read or change calendar events",
                          "email": "send or read email", "search": "search the web"}},
})  # -> calendar

To load it by name (Decider.load("micro-general"), sparrow-decider --model micro-general ...), put the file in ~/.cache/sparrow-decider/ or any folder listed in $SPARROW_MODELS:

hf download sparrow-decider/sparrow-micro-general sparrow-micro-general-v1.1.sparrow --local-dir ~/.cache/sparrow-decider

C

sparrow-micro-general-v1.1.spb is the packed blob for the C99 runtime in c/sparrow. See the C library guide.

#include "sparrow.h"
static sparrow_model m;
static sparrow_ctx ctx;

sparrow_load(&m, blob, blob_len);
float p_true = sparrow_noul(&m, &ctx, "motor_temp_c: 96\nmax_temp_c: 80", "Is the motor overheating?");

The default SPARROW_MAX_DIM (256) fits micro.

Measured results

All numbers are from the project's bench/RESULTS.md and site/facts.json. Local runs on one Apple M4 Pro core, not leaderboard scores. micro has not been benchmarked on a microcontroller.

Speed

Where	Time per decision (p50)
Laptop, 1 core, C runtime	0.22 ms (217 µs)
Laptop, 1 core, numpy	0.38 ms (RobotBench) · 0.6 ms p50 / 4.93 ms p95 (JevBench)
Peak RSS, numpy, JevBench run	99.4 MB

Quality

Benchmark	sparrow-micro-general-v1.1	Strands Decider 2B (public)
MASSIVE 1.1 en-US, 600 held-out utterances: domain / intent	0.635 / 0.542	0.86 / 0.871
JevBench public, 231 tasks (local v1.5 proxy)	117/231
RobotBench v0, 55 hand-written robot decisions	30/55	40/55

C parity: on the host, the C runtime agrees with numpy on 586/586 argmaxes, max probability difference 3e-6.

Limitations

• Static embeddings pool words without order: negation and multi-step reasoning are weak.
• Open-ended language it never trained on is well behind the 2B (MASSIVE routing above).
• Yes/no judgement on JevBench is near chance at every Sparrow tier. A noul near 0.5 means "don't know".
• No key-value number pooling: for decisions that hinge on sensor numbers, use sparrow-nano-robot.
• English only. Text normalisation is exact for Latin scripts.
• It suggests decisions. Never use it as the only safety stop on a physical system.

Training

• Init: potion-base-8M static embeddings, full vocabulary, 256-d.
• Architecture: mean-pooled state / question / option vectors and an MLP that scores every option against the state and the question.
• Data: a general decision mix plus generated robot decisions. Only tasksource rows marked for commercial use were used; sources keep their own licences.
• Teacher: distilled from a Strands Decider 4B checkpoint (targets 50% teacher, 50% gold) on about 430k decisions; per-kind temperature calibration.
• Compression: int8 matrices with a per-row scale.

Details: research report.

Licence and credits

• Licence: Apache-2.0 (see LICENSE and NOTICE in this repo).
• Base weights: derived from potion-base-8M (Model2Vec, MinishLab), released under the MIT licence. Its copyright notice and full MIT text are in LICENSE-potion-base-8M.txt in this repo and apply to the corresponding weights.
• Teacher and prompt format: Strands Decider (Apache-2.0).
• Evaluation only: Amazon MASSIVE 1.1 (CC BY 4.0), JevBench (MIT).

Unofficial community project. Not an AWS or Strands product.

Contact

• Email: vivek@sparrowdecider.com
• Issues: https://github.com/sparrow-decider/sparrow-decider/issues
• Discussions: https://github.com/sparrow-decider/sparrow-decider/discussions

Citation

@software{sparrow_decider,
  author  = {Raja P S, Vivek},
  title   = {Sparrow Decider: decision models from 47 MB down to a few KB},
  year    = {2026},
  version = {0.2.0},
  url     = {https://github.com/sparrow-decider/sparrow-decider}
}