Instructions to use oldman-dev/tis-v8b-hard-anchor with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use oldman-dev/tis-v8b-hard-anchor with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("oldman-dev/tis-v8b-hard-anchor", dtype="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "model": "mistralai/Mistral-7B-v0.3", | |
| "approach": "V8_hard_anchor_restored", | |
| "anchor_force_value": 1.0, | |
| "loss_weights": { | |
| "ranking": 1.0, | |
| "retrieval": 2.0, | |
| "stability": 0.5 | |
| }, | |
| "hyperparameters": { | |
| "steps": 2000, | |
| "lr": 0.001, | |
| "grad_accum": 4, | |
| "margin_hard_neg": 0.5, | |
| "budgets": [ | |
| 0.25, | |
| 0.5, | |
| 0.75 | |
| ] | |
| } | |
| } |