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": "small",
"difficulty": 0.4,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.1,
"is_sensitive": 0.1,
"tier": "small",
"why": "difficulty 0.40 < 1.5, no tools, not sensitive",
"row": " label 0.40 / 3 writing (0.30) 0.100 0.100"
},
{
"name": "hard",
"difficulty": 2.2,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.2,
"is_sensitive": 0.3,
"tier": "frontier",
"why": "difficulty 2.20 >= 1.5",
"row": " label 2.20 / 3 writing (0.30) 0.200 0.300"
},
{
"name": "at-difficulty-threshold",
"difficulty": 1.5,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.2,
"is_sensitive": 0.3,
"tier": "frontier",
"why": "difficulty 1.50 >= 1.5",
"row": " label 1.50 / 3 writing (0.30) 0.200 0.300"
},
{
"name": "just-under-threshold",
"difficulty": 1.4999,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.2,
"is_sensitive": 0.3,
"tier": "small",
"why": "difficulty 1.50 < 1.5, no tools, not sensitive",
"row": " label 1.50 / 3 writing (0.30) 0.200 0.300"
},
{
"name": "tools",
"difficulty": 0.2,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.5,
"is_sensitive": 0.49,
"tier": "frontier+tools",
"why": "needs external data (p=0.50)",
"row": " label 0.20 / 3 writing (0.30) 0.500 0.490"
},
{
"name": "sensitive-overrides-tools",
"difficulty": 0.2,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.9,
"is_sensitive": 0.5,
"tier": "frontier",
"why": "sensitive: money/legal/medical/safety (p=0.50)",
"row": " label 0.20 / 3 writing (0.30) 0.900 0.500"
},
{
"name": "tools-just-under",
"difficulty": 0.2,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.4999,
"is_sensitive": 0.1,
"tier": "small",
"why": "difficulty 0.20 < 1.5, no tools, not sensitive",
"row": " label 0.20 / 3 writing (0.30) 0.500 0.100"
},
{
"name": "tools-overrides-hard",
"difficulty": 2.2,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.9,
"is_sensitive": 0.1,
"tier": "frontier+tools",
"why": "needs external data (p=0.90)",
"row": " label 2.20 / 3 writing (0.30) 0.900 0.100"
},
{
"name": "sensitive-overrides-hard",
"difficulty": 2.5,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.1,
"is_sensitive": 0.9,
"tier": "frontier",
"why": "sensitive: money/legal/medical/safety (p=0.90)",
"row": " label 2.50 / 3 writing (0.30) 0.100 0.900"
},
{
"name": "sensitive-easy",
"difficulty": 0.1,
"domains": {
"code": 0.1,
"math_or_logic": 0.2,
"writing": 0.3,
"factual_lookup": 0.3,
"data_analysis": 0.05,
"chitchat": 0.05
},
"needs_tools": 0.0,
"is_sensitive": 0.97,
"tier": "frontier",
"why": "sensitive: money/legal/medical/safety (p=0.97)",
"row": " label 0.10 / 3 writing (0.30) 0.000 0.970"
}
]