Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
Python calibrationclassificationdecision-modelhuggingfacejevmodernbertmultilingualnlppythonpytorchroutingtyped-decisionszero-shot
{
"name": "nonstring_instructions",
"state": "Le client demande un remboursement immédiat.",
"serialized_state": "Le client demande un remboursement immédiat.",
"questions": {
"structured_ins": {
"type": "choice",
"instructions": {
"task": "route",
"hint": "Müller & co <urgent>",
"n": 2.0
},
"criteria": {
"a": "first",
"b": "second"
}
},
"list_ins": {
"type": "noul",
"instructions": [
"step one",
"step two",
3
]
}
},
"per_question": [
{
"id": "structured_ins",
"type": "choice",
"qtype": 0,
"instructions": "{\"task\": \"route\", \"hint\": \"Müller & co <urgent>\", \"n\": 2.0}",
"head_text": "choice question: {\"task\": \"route\", \"hint\": \"Müller & co <urgent>\", \"n\": 2.0}",
"options": [
"a: first",
"b: second"
],
"input_ids": [
2,
6241,
2872,
235292,
19946,
8277,
1192,
664,
11657,
824,
664,
38089,
1192,
664,
185500,
724,
910,
968,
138437,
42937,
664,
235254,
1192,
235248,
235284,
235265,
235276,
235270,
1,
4,
476,
235292,
1370,
4,
518,
235292,
2257,
1,
1709,
5553,
28355,
748,
130787,
65681,
482,
235265,
1
],
"markers": [
29,
33
]
},
{
"id": "list_ins",
"type": "noul",
"qtype": 2,
"instructions": "[\"step one\", \"step two\", 3]",
"head_text": "noul question: [\"step one\", \"step two\", 3]",
"options": [
"false: no, the statement does not hold",
"true: yes, the statement holds"
],
"input_ids": [
2,
552,
6311,
2872,
235292,
10890,
8847,
974,
824,
664,
8847,
1378,
824,
235248,
235304,
235307,
1,
4,
1566,
235292,
793,
235269,
573,
6218,
1721,
780,
3385,
4,
1382,
235292,
7778,
235269,
573,
6218,
12723,
1,
1709,
5553,
28355,
748,
130787,
65681,
482,
235265,
1
],
"markers": [
17,
27
]
}
],
"batch": {
"input_ids": [
[
2,
6241,
2872,
235292,
19946,
8277,
1192,
664,
11657,
824,
664,
38089,
1192,
664,
185500,
724,
910,
968,
138437,
42937,
664,
235254,
1192,
235248,
235284,
235265,
235276,
235270,
1,
4,
476,
235292,
1370,
4,
518,
235292,
2257,
1,
1709,
5553,
28355,
748,
130787,
65681,
482,
235265,
1
],
[
2,
552,
6311,
2872,
235292,
10890,
8847,
974,
824,
664,
8847,
1378,
824,
235248,
235304,
235307,
1,
4,
1566,
235292,
793,
235269,
573,
6218,
1721,
780,
3385,
4,
1382,
235292,
7778,
235269,
573,
6218,
12723,
1,
1709,
5553,
28355,
748,
130787,
65681,
482,
235265,
1,
0,
0
]
],
"attention_mask": [
[
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1
],
[
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
0,
0
]
],
"marker_pos": [
[
29,
33
],
[
17,
27
]
],
"marker_mask": [
[
true,
true
],
[
true,
true
]
],
"qtype": [
0,
2
]
},
"marker_axis_padded": false,
"logits": [
[
-0.9075420498847961,
-0.7191608548164368
],
[
-1.4057300090789795,
0.24854958057403564
]
],
"act_logits": [
[
1266.0982666015625,
-1506.8525390625
],
[
1164.3819580078125,
-1387.8626708984375
]
],
"result": {
"model": "laya-rl-agent",
"answers": {
"structured_ins": {
"type": "choice",
"choice": "b",
"probabilities": {
"a": 0.453,
"b": 0.547
},
"confidence": 0.0064,
"answer_confidence": 0.547,
"action": {
"act_probability": 1.0
}
},
"list_ins": {
"type": "noul",
"noul": 0.8395,
"confidence": 0.8395,
"answer_confidence": 0.8395,
"action": {
"act_probability": 1.0
}
}
},
"usage": {
"input_tokens": 92,
"output_tokens": 0,
"state_tokens": 8,
"state_tokens_dropped": 0,
"truncated": false,
"truncated_questions": []
}
}
}