YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

NTILC Hierarchical Circle v0.1.0

NTILC tool-embedding checkpoint prepared for OpenToolEmbeddings/ntilc-hierarchical-circle-v0-1-0.

Variant

  • Architecture: hierarchical
  • Loss: circle
  • Encoder: sentence-transformers/all-MiniLM-L6-v2
  • Embedding dimension: 128
  • Dataset version: oss-qwen3.5-27b-26tools-312queries-v0.1.0
  • Synthetic generator: Qwen/Qwen3.5-27B
  • Tool count: 26
  • Query count: 312
  • Checkpoint SHA256: 24b334c2cc674474f91af91b697db58657d51d79ea386a1f1a3207b4b2a01944

Files

  • best.pt: best validation checkpoint bundle
  • metrics.json: per-epoch training and validation metrics

Usage

These are NTILC checkpoint bundles, not plain transformers repositories. Load them with the local helper below.

from training import load_checkpoint_bundle

bundle = load_checkpoint_bundle("best.pt", device="cpu")
print(bundle["architecture"], bundle["loss_name"], len(bundle["tool_names"]))

Metrics

  • epochs: 25
  • best_epoch: 12
  • tool_count: 26
  • query_count: 312
  • best_val_tool_retrieval_accuracy: 0.8269
  • best_val_parent_retrieval_accuracy: 0.9038
  • best_val_tool_classification_accuracy: 0.6154
  • best_val_parent_classification_accuracy: 0.9038
  • tool_silhouette_score: 0.0092
  • parent_silhouette_score: 0.2756
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support