Gjusev
/

MK-SH-1

numeric-ordering

typed-decisions

selective-prediction

Model card Files Files and versions

611 MB

Ctrl+K

Ctrl+K

1 contributor

History: 20 commits

Gjusev's picture

Model card: held-out selective risk per set for the frozen E36 policy (5 % budget missed on OOD extremes and routing task)

23efe74 verified 4 days ago

.gitattributes

1.57 kB
Upload tokenizer.json with huggingface_hub 6 days ago
ARCHITECTURE.md

633 Bytes
Upload ARCHITECTURE.md with huggingface_hub 6 days ago
README.md

10.8 kB
Model card: held-out selective risk per set for the frozen E36 policy (5 % budget missed on OOD extremes and routing task) 4 days ago
inference-policy.json

1.62 kB
policy: weights note states E35 training + E36 calibration, no retraining 5 days ago
logo-mark.png

80.2 kB
logo: final MAREA tile (4-color ship version) 5 days ago
model.safetensors

577 MB
xet

Upload model.safetensors with huggingface_hub 6 days ago
tokenizer.json

34.4 MB
xet

Upload tokenizer.json with huggingface_hub 6 days ago
tokenizer_config.json

599 Bytes
Upload tokenizer_config.json with huggingface_hub 6 days ago