nagisanzeninz

Nagi-ENORMOUS

README

License: apache-2.0

Public four-system decision suite

Same protocol and scoring as the Nagi-HUGE release. Typed-choice adaptations of 8 public sources, 4,671 core examples, each source weighted equally.

Table with columns: System, Same complete evidence (4,518), Full operational suite (4,671)
System	Same complete evidence (4,518)	Full operational suite (4,671)
Jev1.13.0	83.95%	83.92%
Nagi-ENORMOUS	82.76%	83.07%
Nagi-HUGE	77.49%	77.92%
OpenJev / SemIf	73.87%	74.06%
Laya	57.35%	57.25%

ENORMOUS − Jev: −1.20 pp [−2.70, +0.31]. The interval includes zero, so the two are statistically tied on this suite.
ENORMOUS − HUGE: +5.27 pp [+3.49, +7.14].
By source, ENORMOUS is ahead of Jev on XNLI in all four languages (en 93.5, vi 83.0, de 86.0, zh 79.5), ANLI, BoolQ, CB and RTE, and behind on MMLU-Pro (65.0 vs 82.3).
The option-order flip rate was not measured for ENORMOUS.
This is not an official leaderboard score, and exposure to these sources during pretraining is unknown.

Arena Live: real-time games

We ran three games with real rules and physics. Every model got the same observations and options, moves were simultaneous, seeds were pre-committed, seats rotated and play was lockstep. Each game had 10 rounds, scored 3/2/1/0 points per round (max 90 points, 30 round wins).

Table with columns: Model, Total points, Round wins, Rotorwash / Lightcycle / Stack points
Model	Total points	Round wins	Rotorwash / Lightcycle / Stack points
Nagi-ENORMOUS 27B	83 / 90	23 / 30	25 / 28 / 30
Jev 1.13.0	57 / 90	5 / 30	25 / 12 / 20
OpenJev (SemIf 4B)	20 / 90	2 / 30	0 / 20 / 0
Laya	20 / 90	0 / 30

Rotorwash (2D rigid-body helicopter): ENORMOUS and Jev tie on points, but ENORMOUS stayed airborne longer in 9/10 rounds (median 35.8 s vs 13.8 s).
Lightcycle Royale: ENORMOUS won 8 of 10 rounds.
Stack Attack: ENORMOUS won 10 of 10 rounds.
Among Nagi lines: 7 → 39 → 65 → 76 points for SMOL → BIG → HUGE → ENORMOUS.
Scope: exhibition scale (10 rounds per game); every model got the same shared sensor features. Arena Live measures skill under fixed, known rules. On our internal test of reading brand-new rules, the 27B line does not yet improve over the 12B line.

Videos: https://nagisanzenin.github.io/nagi/arena-live/

Use

bash

python -m pip install "nagi-decisions @ git+https://github.com/nagisanzenin/nagi.git"

python

from nagi import load_enormousmodel = load_enormous(device="cuda")result = model.system_one(    state={"A": 7, "B": 3},    questions={"decision": {"type": "choice", "instructions": "Which value is greater?",                            "criteria": {"A": "A is greater", "B": "B is greater"}}},)print(result["answers"]["decision"])

Validated path: unmerged LoRA, CUDA BF16, eager attention, the official Qwen chat template with enable_thinking=False, and selected option-letter projection.
Temperature: 1.0, which is not a fitted calibration.
Options and length: choice supports 2–26 options. Training prompts were ≤768 tokens; longer inputs are outside the trained range. Rendered input is capped at 4,096 tokens and fails with an explicit error, with no silent truncation.
Memory: the BF16 weights need about 56 GB, so plan for an 80 GB GPU.

Training and selection

Recipe: HUGE's. A 75-update pilot, then continued supervised LoRA (effective batch 8, AdamW learning rate 3e−5), rank 8, alpha 16, no dropout.
Selection: release rungs at 125, 250 and 500. The preregistered stop rule ended training at step 500 (dev policy 0.9397, language 0.8631), and step 500 was selected on dev only.
Evaluation data: no public evaluation gold and no final-set data were used for selection.
Adapter weights SHA256: b8d307527a0426fd8208bf3c44eebaf47f14057452403ee282c08fe72053ddf7
Adapter config SHA256: 9b9529fcf29318ff348abfd2db9face3ecdd0b9bf281734cec739ec186d86c09

Links

SDK and documentation · Nagi-HUGE · Nagi-BIG · Nagi-SMOL

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

nagisanzeninz

Model Tree

Base

Qwen/Qwen3.8-27B

Adapter

this model

Input Modalities

TextImageVideo

Output Modalities

Text

Supported Functionality