f795aa5604
Fix survival calculation in cifs_from_exponential_logits: broadcast mask for compatibility with tensor shapes
2026-01-10 11:42:03 +08:00
f231a2e4e5
Add model evaluation configurations for Delphi and SAP Delphi models
...
- Introduced a new JSON file `models_eval_example.json` containing evaluation configurations for various models.
- Included configurations for `delphi_fork` and `sap_delphi` models with different loss types and full covariance settings.
- Each model entry specifies the model name, type, loss type, full covariance flag, and checkpoint path.
2026-01-10 11:37:12 +08:00
3f15301f26
Enhance multi-GPU experiment runner: add log directory option, improve logging with sanitized filenames, and capture command output in log files.
2026-01-09 18:46:59 +08:00
5f8360a24a
Update experiments configuration: replace PiecewiseExponentialLoss with DiscreteTimeCIFNLLLoss for delphi_fork and sap_delphi models
2026-01-09 18:44:16 +08:00
739eb326f2
Fix formatting in DiscreteTimeCIFNLLLoss documentation for clarity
2026-01-09 18:34:01 +08:00
209dde2299
Refactor loss functions and model architecture: replace PiecewiseExponentialLoss with DiscreteTimeCIFNLLLoss, update Trainer to use SimpleHead, and modify argument parsing for new loss type.
2026-01-09 18:31:38 +08:00
880fd53a4b
Enhance Trainer class: add logging for WeibullNLLLoss parameters during training and validation
2026-01-09 13:48:36 +08:00
8723bf7600
Enhance Trainer class: add delta_scale logging for improved training statistics
2026-01-09 13:28:11 +08:00
dc34d51864
Add rank parameter to TrainConfig and update argument parsing for low-rank parameterization
2026-01-09 13:18:09 +08:00
1fa6d55d79
Refactor PiecewiseExponentialLoss and WeibullNLLLoss: remove lightweight numerical protections and improve error handling for input validation
2026-01-09 13:06:43 +08:00
b54c54a60b
Refactor Trainer class: improve training statistics calculation and logging for NLL and regularization
2026-01-09 12:49:29 +08:00
aff0fe480b
Refactor model and training scripts: remove unused imports and add FactorizedHead class for improved modularity
2026-01-09 12:01:52 +08:00
c70c3cd71e
Reorganize import statements for consistency and clarity in model and training scripts
2026-01-09 10:16:03 +08:00
d7600514af
Update Trainer class to rename parameters for clarity in embedding configuration
2026-01-08 17:11:19 +08:00
d20d32ba22
Add multi-GPU experiment runner script and experiments configuration
2026-01-08 13:57:34 +08:00
01a96d37ea
Enhance DataLoader configuration and improve tensor transfer efficiency in Trainer class
2026-01-08 13:20:32 +08:00
5382f9f159
Remove EMA model implementation from Trainer class and related parameters from TrainConfig
2026-01-08 13:14:29 +08:00
615e2fe748
Refactor PiecewiseExponentialLoss for clarity and numerical stability improvements
2026-01-08 13:05:53 +08:00
06a01d2893
Add PiecewiseExponentialLoss class and update TrainConfig for new loss type
2026-01-08 12:45:31 +08:00
7c36f7a007
Update age_encoder parameter choices in TrainConfig and argument parser for clarity
2026-01-08 11:38:45 +08:00
9eda00ea48
Add n_tech_tokens parameter to DelphiFork and SapDelphi model initializations
2026-01-08 11:36:23 +08:00
1d1f568a3f
Rename age_encoder parameter to age_encoder_type for clarity in Trainer class
2026-01-08 11:34:44 +08:00
8293f7ee24
Enhance TabularEncoder with BatchNorm and AutoDiscretization for continuous features
2026-01-08 00:24:43 +08:00
33ba7e6c1d
Refactor training logic to improve early stopping mechanism and variable naming
2026-01-08 00:07:15 +08:00
811b2e1a46
Implement training script for Delphi model with configurable parameters
2026-01-07 23:57:29 +08:00
6984b254b3
Add loss functions and model architecture for time-to-event prediction
...
- Implemented ExponentialNLLLoss and WeibullNLLLoss in losses.py for negative log-likelihood calculations.
- Developed TabularEncoder class in model.py for encoding tabular features.
- Created DelphiFork and SapDelphi classes in model.py for time-to-event prediction using transformer architecture.
- Added data preparation scripts in prepare_data.R and prepare_data.py for processing UK Biobank data, including handling field mappings and event data extraction.
2026-01-07 21:32:00 +08:00
5d1d79b908
Initial commit
2026-01-07 21:27:38 +08:00
