From 178772b81937dddc67573ad557a2f3573fde9292 Mon Sep 17 00:00:00 2001 From: siqube Date: Mon, 6 Apr 2026 14:50:43 +0200 Subject: [PATCH] fix constant names --- .github/copilot-instructions.md | 6 +++--- README.md | 5 ++--- docs/index.md | 2 +- 3 files changed, 6 insertions(+), 7 deletions(-) diff --git a/.github/copilot-instructions.md b/.github/copilot-instructions.md index 1bf47037..d7cac3ab 100644 --- a/.github/copilot-instructions.md +++ b/.github/copilot-instructions.md @@ -41,7 +41,7 @@ python src/run/single_run/train.py +trainer=TrainerDL +model=RoberteyeWord +data ``` #### Multi-Run Pipeline (Production) -1. **Model checker**: `src/run/multi_run/model_checker.sh` - runs train+test for all model/data combinations +1. **Model checker**: `run_commands/utils/model_checker.sh` - runs train+test for all model/data combinations 2. **Sweep generation**: `src/run/multi_run/sweep_wrapper.sh` - creates hyperparameter sweeps 3. **Results processing**: `raw_to_processed_results.py` → `csv_to_latex.py` for paper tables @@ -80,7 +80,7 @@ bash src/data/preprocessing/get_data.sh DATASET1,DATASET2 # Specific datasets ### Model Training/Testing Pipeline ```bash # Quick model validation (recommended for development) -bash src/run/multi_run/model_checker.sh --data_tasks CopCo_TYP --folds 0,1 --cuda 0 +bash run_commands/utils/model_checker.sh --data_tasks CopCo_TYP --folds 0,1 --cuda 0 # Full evaluation pipeline bash src/run/multi_run/test_wrapper_creator.sh # Generates test scripts @@ -109,4 +109,4 @@ python src/run/multi_run/csv_to_latex.py # Generate paper tables - `results/`: Raw trial-level results and processed benchmark tables - `logs/`: Completed/failed run tracking for multi-run experiments -When implementing features, follow the existing factory pattern and registration decorators. The codebase emphasizes reproducibility and cross-validation rigor for eye movement research. \ No newline at end of file +When implementing features, follow the existing factory pattern and registration decorators. The codebase emphasizes reproducibility and cross-validation rigor for eye movement research. diff --git a/README.md b/README.md index 7df0b1fe..65888ca7 100644 --- a/README.md +++ b/README.md @@ -169,9 +169,8 @@ results/eyebench_benchmark_results/{metric}.csv Implement `forward()` and `shared_step()` methods. 2. Register it in: - - `src/configs/enums.py` → `ModelNames` - - `src/configs/model_args.py` → model config class - - `src/configs/config.py` → `ModelMapping` + - `src/configs/constants.py` → `DLModelNames` (or `MLModelNames` for ML models) + - `src/configs/models/dl/YourModel.py` → model config class decorated with `@register_model_config` (use `src/configs/models/ml/` for ML models) 3. Define its default parameters and search space in `src/run/multi_run/search_spaces.py`. 4. Verify integration: diff --git a/docs/index.md b/docs/index.md index 0b124897..a1d7e322 100644 --- a/docs/index.md +++ b/docs/index.md @@ -175,7 +175,7 @@ results/eyebench_benchmark_results/{metric}.csv 4. Verify integration: ```bash -bash src/run/multi_run/model_checker.sh +bash run_commands/utils/model_checker.sh ``` ---