Instructions to use Ankit1106/laya-typed-decisions with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Ankit1106/laya-typed-decisions with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Ankit1106/laya-typed-decisions", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download laya_benchmark_report.json from Ankit1106/laya-typed-decisions: direct link, hf CLI and curl.
- Browser
- Download file 774 Bytes
-
https://huggingface.co/Ankit1106/laya-typed-decisions/resolve/main/laya_benchmark_report.json
- Command line
-
hf download hf://Ankit1106/laya-typed-decisions/laya_benchmark_report.json
-
curl -L -o laya_benchmark_report.json https://huggingface.co/Ankit1106/laya-typed-decisions/resolve/main/laya_benchmark_report.json
774 Bytes
| { | |
| "benchmark": "LocalLLaMA/typed-decisions", | |
| "model": "Laya (Fine-Tuned 2xT4)", | |
| "n_cases": 400, | |
| "n_decisions": 2000, | |
| "metrics": { | |
| "accuracy": 0.769, | |
| "soft_accuracy": 0.5068, | |
| "brier_score": 0.0694, | |
| "ece": 0.215, | |
| "score_mae": 0.2442, | |
| "within_1_level": 0.9912, | |
| "latency_p50_ms": 116.1, | |
| "latency_p95_ms": 153.8, | |
| "kl_divergence": 0.1287, | |
| "total_variation": 0.184 | |
| }, | |
| "per_workflow": { | |
| "agent_trace_observability": { | |
| "decisions": 500, | |
| "accuracy": 0.746 | |
| }, | |
| "customer_service": { | |
| "decisions": 500, | |
| "accuracy": 0.776 | |
| }, | |
| "invoice_processing": { | |
| "decisions": 500, | |
| "accuracy": 0.806 | |
| }, | |
| "security_incidents": { | |
| "decisions": 500, | |
| "accuracy": 0.748 | |
| } | |
| } | |
| } |