Instructions to use JonasGeiping/crammed-bert-legacy with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use JonasGeiping/crammed-bert-legacy with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="JonasGeiping/crammed-bert-legacy")# Load model directly from transformers import AutoModelForPreTraining model = AutoModelForPreTraining.from_pretrained("JonasGeiping/crammed-bert-legacy", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "optim": { | |
| "type": "AdamW", | |
| "lr": 0.001, | |
| "betas": [ | |
| 0.9, | |
| 0.98 | |
| ], | |
| "eps": 1e-12, | |
| "weight_decay": 0.01, | |
| "amsgrad": false | |
| }, | |
| "optim_mod": { | |
| "name": "none" | |
| }, | |
| "name": "bert-o3", | |
| "limited_decay_keys": [ | |
| "bias", | |
| "LayerNorm.bias", | |
| "LayerNorm.weight" | |
| ], | |
| "warmup_steps": 0, | |
| "cooldown_steps": 0, | |
| "steps": 600000, | |
| "scheduler": "budget-triangle2", | |
| "batch_size": 4096, | |
| "batch_size_ramp": 300000, | |
| "gradient_clipping": 0.5, | |
| "pretrain_in_train_mode": false, | |
| "objective": { | |
| "name": "masked-lm", | |
| "mlm_probability": 0.15, | |
| "use_80_20_rule": true, | |
| "disable_mlm": false, | |
| "token_drop": 0.0 | |
| }, | |
| "reverse_dataset_order": false, | |
| "budget": 24, | |
| "gradinit": { | |
| "enabled": false | |
| } | |
| } |