MLX
Joblib
Safetensors
English
reasoning
chain-of-thought
context-compression
soft-prompt
apple-silicon
Instructions to use baya1116/hypernet-sp-distill with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use baya1116/hypernet-sp-distill with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir hypernet-sp-distill baya1116/hypernet-sp-distill
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
| [data] train 150 held 80 | |
| [data] gate pts tr 40334(5762+) te 22589(3227+) idx te 3227 | |
| [A] gate AUC=0.9537 best-F1={'f1': 0.725, 'precision': 0.674, 'recall': 0.785, 'thresh': 0.696} R@P0.9=0.359 shipped={'thresh': 4.04, 'precision': 0.866, 'recall': 0.469} | |
| [B] gate AUC logistic=0.9500 mlp256=0.9607 mlp512=0.9604 | |
| [C] backward-scan {"tau=-32.12": {"reach_rate": 0.755, "avg_blocks_scanned": 7.18}, "tau=-31.70": {"reach_rate": 0.616, "avg_blocks_scanned": 7.22}, "tau=-31.32": {"reach_rate": 0.445, "avg_blocks_scanned": 7.22}, "gold_recency_rank": {"mean": 6.49, "median": 7, "frac_in_recent_3": 0.001, "frac_in_recent_8": 0.747}} | |
| === EXPLORE SUMMARY === | |
| { | |
| "data": { | |
| "gate_pts_held": 22589, | |
| "pos_held": 3227, | |
| "idx_held": 3227 | |
| }, | |
| "A_gate_operating_point": { | |
| "auc": 0.9537, | |
| "best_f1": { | |
| "f1": 0.725, | |
| "precision": 0.674, | |
| "recall": 0.785, | |
| "thresh": 0.696 | |
| }, | |
| "recall_at_p0.9": 0.359, | |
| "precision_at_r0.9": 0.526, | |
| "shipped_threshold": { | |
| "thresh": 4.04, | |
| "precision": 0.866, | |
| "recall": 0.469 | |
| } | |
| }, | |
| "B_gate_model_search": { | |
| "logistic_auc": 0.95, | |
| "mlp256_auc": 0.9607, | |
| "mlp512_auc": 0.9604 | |
| }, | |
| "C_backward_scan": { | |
| "tau=-32.12": { | |
| "reach_rate": 0.755, | |
| "avg_blocks_scanned": 7.18 | |
| }, | |
| "tau=-31.70": { | |
| "reach_rate": 0.616, | |
| "avg_blocks_scanned": 7.22 | |
| }, | |
| "tau=-31.32": { | |
| "reach_rate": 0.445, | |
| "avg_blocks_scanned": 7.22 | |
| }, | |
| "gold_recency_rank": { | |
| "mean": 6.49, | |
| "median": 7, | |
| "frac_in_recent_3": 0.001, | |
| "frac_in_recent_8": 0.747 | |
| } | |
| }, | |
| "D_bge_small": { | |
| "note": "skipped (block texts not in IX rows); base bge_head is canonical" | |
| }, | |
| "final": true, | |
| "elapsed_s": 340.4 | |
| } | |
| DONE t=340s |