Text Classification
Transformers
ONNX
Safetensors
Rust
English
tool-use
function-calling
tool-calling
guardrails
verifier
shadow-mode
Instructions to use cowWhySo/toolcall-verifier-classifier-production with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use cowWhySo/toolcall-verifier-classifier-production with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="cowWhySo/toolcall-verifier-classifier-production")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("cowWhySo/toolcall-verifier-classifier-production", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "schema_version": "toolcall-verifier-candidate-thresholds/v1", | |
| "source": "validation_calibration_diagnostics_only", | |
| "temperature": 1.1008745431900024, | |
| "deployment_rule": { | |
| "nonvalid_confidence_thresholds": { | |
| "wrong_tool_semantic": 0.98, | |
| "wrong_arguments_semantic": 0.98, | |
| "tool_not_needed": 0.98, | |
| "needs_clarification": 0.99 | |
| }, | |
| "min_margin_over_valid": 0.25, | |
| "valid_logit_bias": 0.0, | |
| "valid_false_objection_gate": 0.005 | |
| }, | |
| "recommended_action_policy": { | |
| "mode": "advisory", | |
| "deployment_thresholds": { | |
| "wrong_tool_semantic": 0.98, | |
| "wrong_arguments_semantic": 0.98, | |
| "tool_not_needed": 0.98, | |
| "needs_clarification": 0.99 | |
| }, | |
| "min_margin_over_valid": 0.25, | |
| "valid_logit_bias": 0.0, | |
| "raw_0_90_false_objection": "diagnostic_only", | |
| "enforce": false | |
| }, | |
| "notes": [ | |
| "Candidate thresholds are diagnostics only.", | |
| "The active artifact mode remains shadow; the recommended next mode is advisory after replay.", | |
| "Promotion is evaluated against the guarded deployment rule, not raw top-label confidence alone." | |
| ], | |
| "labels": { | |
| "valid": { | |
| "action": "allow", | |
| "advisory_min_confidence": 0.0, | |
| "enforce_min_confidence": 1.01 | |
| }, | |
| "wrong_tool_semantic": { | |
| "action": "diagnostic_only", | |
| "advisory_min_confidence": 0.8863924050632912, | |
| "enforce_min_confidence": 0.95, | |
| "deployment_candidate_min_confidence": 0.98, | |
| "deployment_min_margin_over_valid": 0.25, | |
| "deployment_valid_logit_bias": 0.0 | |
| }, | |
| "wrong_arguments_semantic": { | |
| "action": "diagnostic_only", | |
| "advisory_min_confidence": 1.01, | |
| "enforce_min_confidence": 1.01, | |
| "deployment_candidate_min_confidence": 0.98, | |
| "deployment_min_margin_over_valid": 0.25, | |
| "deployment_valid_logit_bias": 0.0 | |
| }, | |
| "tool_not_needed": { | |
| "action": "diagnostic_only", | |
| "advisory_min_confidence": 0.9579746835443038, | |
| "enforce_min_confidence": 0.9568354430379746, | |
| "deployment_candidate_min_confidence": 0.98, | |
| "deployment_min_margin_over_valid": 0.25, | |
| "deployment_valid_logit_bias": 0.0 | |
| }, | |
| "needs_clarification": { | |
| "action": "diagnostic_only", | |
| "advisory_min_confidence": 0.8765189873417721, | |
| "enforce_min_confidence": 0.95, | |
| "deployment_candidate_min_confidence": 0.99, | |
| "deployment_min_margin_over_valid": 0.25, | |
| "deployment_valid_logit_bias": 0.0 | |
| }, | |
| "deterministic_invalid": { | |
| "action": "deterministic_only", | |
| "advisory_min_confidence": 1.01, | |
| "enforce_min_confidence": 1.01 | |
| } | |
| }, | |
| "promotion_status": "promotable_pending_replay", | |
| "blocked_reasons": [], | |
| "artifact_promotable": true | |
| } |