Downloads · 30 days
0
nagisanzeninz/nagi-smol-v0
nagi-smol-v0 is a machine learning model from nagisanzeninz. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as apache-2.0.
System One typed-decision model (choice / score / noul) — M2′ dual encoder (ModernBERT-large 421M) + option tower. Independent per-option encoding + bilinear MLP score head. Soft probability labels. No text generation.
Downloads · 30 days
0
Access
Public
Updated Sep 23, 2026
Repo size
78.7 GB
Likes
0
Public
Click a slice to open those files.
.pt78.7 GB · 100%
From the Hugging Face model README
System One typed-decision model (choice / score / noul) — M2′ dual encoder (ModernBERT-large 421M) + option tower. Independent per-option encoding + bilinear MLP score head. Soft probability labels. No text generation.
Also known as: Track-S cycle-2 / hail_s_c2 / nagi-t4s-c2.
import torch
from nagi.config import load_config
from nagi.model import build_model
from nagi.serve.predict import Nagi
cfg = load_config("config.snapshot.yaml") # shipped in this repo
model = build_model(cfg)
state = torch.load("model.pt", map_location="cpu", weights_only=True)
model.load_state_dict(state, strict=False)
nagi = Nagi(model) # .system_one(state, questions) -> answers with calibrated probs
Source: nagisanzenin/nagi (src/nagi/serve/predict.py).
| Nagi-Smol | Nagi-Big | Laya | Jev | |
|---|---|---|---|---|
| LTO hard | 0.462 | 0.709 | ~0.35 | 0.773 |
| ECE raw | 0.070 | — | 0.081 | 0.121 |
| p50 latency | 11 ms (H100) | 84 ms (L4) | 33 ms | 501 ms (API) |
Beats Laya on LTO + latency. 45× faster than Jev API. Honest gap vs Jev on LTO.
render_state_context = definition + 2 demos + state + question