linne-sprout-1.0

linne-sprout-1.0 is a small decision model built on Gemma 4 E4B (4B effective parameters). Give it a state (text, a document, JSON, a conversation or an agent trajectory) and a question with a set of options, and it returns one option together with a probability for every option, in a single forward pass.

It is the smallest member of the linne family: linne-oak-1.0 (31B) and linne-fern-1.0 (27B).

Highlights

• Decision Index public score 48.88, measured with the official kit (edition 0.3, all public benchmarks, 140,620 requests, complete).
• Reads the whole prompt at once: the full-attention layers attend in both directions over the prompt (sliding-window layers keep their native behavior), so the decision sees the state, the question and every option together.
• Strongest on tools and automation (70.3) and a step-level guard for agents with a low false-block rate.
• Small and fast: one forward pass per decision; median 80 ms per request in the official run.

linne-sprout-1.0 is a decision-only model. Because its full-attention layers are bidirectional, it does not generate text: use it to pick options and read probabilities, not as a chat model.

Decision Index (public part, edition 0.3)

Chance-corrected skill, higher is better. linne rows measured by us with the official kit.

Only the public part (20% of the full leaderboard score) can be measured outside the leaderboard.

Usage

The prompt is one user message (chat template, thinking off): the state (compact JSON when not a string), a blank line, Question: <instructions>, one line <label>: <name> - <description> per option, and Answer with the letter of one option only. The answer is the softmax over the next-token logits of the option letters (with the model's final logit soft-capping), read at the last prompt position, with the full-attention layers bidirectional.

linne_engine.py (with nc_hook.py) in this repository implements exactly this, as a Decision Index engine and as a library:

import sys
from huggingface_hub import snapshot_download

path = snapshot_download("MISA-core/linne-sprout-1-0")
sys.path.insert(0, path)
from linne_engine import LinneEngine

model = LinneEngine(model=path, device="cuda:0", noncausal=True)
result, _ = model(
    "My Stripe integration keeps failing. Please help ASAP.",
    {"team": {"type": "choice", "instructions": "Which team should handle this request?",
              "criteria": {"billing": "Charges and refunds", "technical_support": "Integration errors",
                           "sales": "Questions about buying a product"}}},
)
print(result["answers"]["team"])  # chosen option and a probability for every option

Run the Decision Index public suite:

PYTHONPATH=<path> python -m decision_index pipeline --engine linne_engine:LinneEngine \
  --option model=<path> --option device=cuda:0 --option noncausal=1 --out runs/linne-sprout-1-0

BF16 weights are about 16 GB and fit one 24 GB GPU.

License

Apache 2.0, following the base model's license.