Instructions to use Angshul/higher-order-pruner with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Angshul/higher-order-pruner with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Angshul/higher-order-pruner", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download pruner_config.json from Angshul/higher-order-pruner: direct link, hf CLI and curl.
- Browser
- Download file 332 Bytes
-
https://huggingface.co/Angshul/higher-order-pruner/resolve/main/pruner_config.json
- Command line
-
hf download hf://Angshul/higher-order-pruner/pruner_config.json
-
curl -L -o pruner_config.json https://huggingface.co/Angshul/higher-order-pruner/resolve/main/pruner_config.json
332 Bytes
| { | |
| "format": "higher-order-pruner-v1", | |
| "method": "diagonal-higher-order-kfac", | |
| "defaults": { | |
| "sparsity": 0.5, | |
| "probes": 2, | |
| "finite_difference_relative_step": 0.02, | |
| "cubic_cap": 0.5, | |
| "damping_ratio": 0.01, | |
| "trust_ratio": 0.01, | |
| "backtracking": [1.0, 0.5, 0.25, 0.125, 0.0625], | |
| "seed": 91000 | |
| } | |
| } | |