Description
AutoTrust-JEV-27B v1.1 is a sophisticated text decision model hosted on Hugging Face. It is engineered to assign probabilities to answer options provided with each question, facilitating structured decision-making processes. The model's performance is evaluated using the Decision Index 0.3 public suite, where it achieved a public index score of 67.31 across 37 indexed benchmarks and completed 140,178 scoreable requests.
Read More
The model demonstrates significant strengths in tool selection, intent classification, and everyday automation decisions, with Tools & Automation being its highest-scoring area at 85.19 skill points. It excels in specific capabilities such as intent classification, tool selection, banking intent classification, appliance-control decisions, API-use decisions, claim verification, clinical textual inference, and contract inference.
AutoTrust-JEV-27B v1.1's benchmark highlights include impressive scores in intent classification with a 97.06 score on the CLINC150+OOS benchmark, tool selection with a 96.25 score on the BFCL benchmark, and banking intent classification with a 93.63 score on the BANKING77 benchmark. These scores reflect the model's ability to handle complex decision-making tasks effectively.
Despite its strengths, the model's performance in areas like humor preference and consensus ranking remains limited, indicating potential areas for improvement. Users are advised to validate the model on representative application data before relying on its probabilities or decisions, as the public evaluation alone does not establish probability calibration or suitability for high-stakes use.
The model is released under the Apache License 2.0, ensuring open access and collaboration. It is served with vLLM and queried through a decision engine, making it accessible for integration into various applications. AutoTrust-JEV-27B v1.1 is a valuable tool for industries requiring structured decision-making capabilities.
- CategoryAI Models & LLMs
- Pricing ModelOpen source
- DeploymentCloud
- PlatformsWeb
AutoTrust-JEV-27B v1.1 Highlights
Text decision model
Probability assignment
Intent classification
Tool selection
Automation decisions
Claim verification
Clinical textual inference
Contract inference
Getting Started with AutoTrust-JEV-27B v1.1
Access model: Visit Hugging Face
Authenticate: Use API key
Set up environment: Configure local hardware
Integrate via API: Use decision engine
Optimise: Validate with application data
AutoTrust-JEV-27B v1.1's Use Cases
- Intent Classification
- Tool Selection
- Automation Decisions
- Claim Verification
- Contract Inference






