| { | |
| "source_path": "cross_lingual_transfer/logs/train/runs/2025-09-18_21-53-03/checkpoints/steps/step-step=015000.ckpt", | |
| "file_size_mb": 852.5924263000488, | |
| "analysis_time": "2025-11-17T23:44:36.698833", | |
| "global_step": 15000, | |
| "epoch": 0, | |
| "lightning_version": "2.5.1.post0", | |
| "step_from_filename": 15000, | |
| "learning_rate": 0.001, | |
| "model_name": "de-transfer-continued-pretraining-15k", | |
| "batch_size": null, | |
| "run_info": { | |
| "run_name": "de-transfer-continued-pretraining", | |
| "run_type": "crosslingual-transfer", | |
| "base_model": "t5-base", | |
| "base_model_hf": "GermanT5/t5-efficient-gc4-german-base-nl36", | |
| "learning_rate": 0.001, | |
| "gradient_clip": 1.0, | |
| "scheduler": "inverse_sqrt", | |
| "warmup_steps": 1500, | |
| "config_name": "crosslingual_de_transfer.yaml", | |
| "full_name_template": "de-transfer-continued-pretraining-{steps}k" | |
| }, | |
| "base_model": "t5-base", | |
| "training_type": "continued_pretraining", | |
| "gradient_clip": 1.0, | |
| "scheduler": "inverse_sqrt", | |
| "warmup_steps": 1500, | |
| "config_name": "crosslingual_de_transfer.yaml", | |
| "conversion_time": "2025-11-17T23:44:58.612779", | |
| "output_path": "converted_models/hf_release/de-t5-sci-transfer-15k/crosslingual-transfer-run-models/de-transfer-continued-pretraining-15k", | |
| "base_model_id": "GermanT5/t5-efficient-gc4-german-base-nl36", | |
| "base_model_used": "/netscratch/nrauscher/projects/BA-hydra/data/huggingface_cache/hub/models--GermanT5--t5-efficient-gc4-german-base-nl36/snapshots/471c7616ac14cb86f0c177716315a92c26b57504", | |
| "missing_keys": 504, | |
| "unexpected_keys": 0 | |
| } |