Model renaming

.gitattributes

@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 data/examples/Our[[:space:]]data.xlsx filter=lfs diff=lfs merge=lfs -text
 models/EpImAge/datamodule.sav filter=lfs diff=lfs merge=lfs -text
+*.sav filter=lfs diff=lfs merge=lfs -text

app.py

@@ -23,17 +23,17 @@ if Path(dir_out).exists():
     shutil.rmtree(Path(dir_out))
 Path(dir_out).mkdir(parents=True, exist_ok=True)
 
-df_imms = pd.read_excel(f"{dir_root}/models/Immunomarkers/Immunomarkers.xlsx", index_col='feature')
+df_imms = pd.read_excel(f"{dir_root}/models/InflammatoryMarkers/InflammatoryMarkers.xlsx", index_col='feature')
 imms = df_imms.index.values
 imms_log = [f"{f}_log" for f in imms]
-cpgs = pd.read_excel(f"{dir_root}/models/Immunomarkers/CpGs.xlsx", index_col=0).index.values
+cpgs = pd.read_excel(f"{dir_root}/models/InflammatoryMarkers/CpGs.xlsx", index_col=0).index.values
 
 models_imms = {}
 for imm in (pbar := tqdm(imms)):
     pbar.set_description(f"Loading model for {imm}")
-    models_imms[imm] = TabularModel.load_model(f"{dir_root}/models/Immunomarkers/{imm}")
+    models_imms[imm] = TabularModel.load_model(f"{dir_root}/models/InflammatoryMarkers/{imm}")
 
-model_age = TabularModel.load_model(f"{dir_root}/models/EpImAge")
+model_age = TabularModel.load_model(f"{dir_root}/models/EpInflammAge")
 
 bkgrd_xai = pd.read_pickle(f"{dir_root}/models/background-xai.pkl")
 bkgrd_imp = pd.read_pickle(f"{dir_root}/models/background-imputation.pkl")
@@ -60,7 +60,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
     
     gr.Markdown(
         """
-        # EpImAge Calculator
+        # EpInflammAge Calculator
         ## Submit epigenetics data
         """
     )
@@ -70,12 +70,12 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
             gr.Markdown(
                 """
                 ### Instruction
-                - Upload your methylation data for 2228 CpGs from [File](https://github.com/GillianGrayson/EpImAge/tree/main/data/CpGs.xlsx).
+                - Upload your methylation data for 2228 CpGs from [File](https://github.com/GillianGrayson/EpInflammAge/tree/main/data/CpGs.xlsx).
                 - The first column must be a sample ID.
                 - Your data must contain `Age` column for metrics (MAE and Pearson Rho) and Age Acceleration calculation.
                 - Missing values should be `NA` in the corresponding cells.
                 - [Imputation](https://scikit-learn.org/stable/modules/impute.html) of missing values can be performed using KNN, Mean, and Median methods with all methylation data from the [Paper]().
-                - Data example for GSE87571: [File](https://github.com/GillianGrayson/EpImAge/tree/main/data/examples/GSE87571.xlsx).
+                - Data example for GSE87571: [File](https://github.com/GillianGrayson/EpInflammAge/tree/main/data/examples/GSE87571.xlsx).
                 - Calculations take a few minutes, the plot can be displayed slightly later than the results. If imputation is performed, the calculations will take longer.
                 """,
             )
@@ -98,12 +98,12 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
     
     with gr.Row():
         with gr.Column():
-            shap_dropdown = gr.Dropdown(label='Choose a sample to get an explanation of the EpImAge prediction', filterable=True, visible=False)
+            shap_dropdown = gr.Dropdown(label='Choose a sample to get an explanation of the EpInflammAge prediction', filterable=True, visible=False)
             shap_button = gr.Button("Get explanation", variant="primary", visible=False)
             with gr.Row():
                 shap_text_id = gr.Text(label='Sample', visible=False)
                 shap_text_age = gr.Text(label='Age', visible=False)
-                shap_text_epimage = gr.Text(label='EpImAge', visible=False)
+                shap_text_epimage = gr.Text(label='EpInflammAge', visible=False)
             shap_markdown_cytokines = gr.Markdown(
                 """
                 ### Most important cytokines:
@@ -189,7 +189,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
         data = pd.read_pickle(f"{dir_out}/{str(request.session_hash)}/data.pkl")
         trgt_id = input
         trgt_age = data.at[trgt_id, 'Age']
-        trgt_pred = data.at[trgt_id, 'EpImAge']
+        trgt_pred = data.at[trgt_id, 'EpInflammAge']
         trgt_aa = trgt_pred - trgt_age
         
         n_closest = 200
@@ -210,7 +210,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
         fig = make_subplots(rows=1, cols=2, shared_yaxes=True, shared_xaxes=False, column_widths=[2.5, 1], horizontal_spacing=0.05, row_titles=[''])
         fig.add_trace(
             go.Waterfall(
-                hovertext=["Chrono Age", "EpImAge"],
+                hovertext=["Chrono Age", "EpInflammAge"],
                 orientation="h",
                 measure=['absolute', 'relative'],
                 y=[-1.5, df_shap.shape[0] + 0.5],
@@ -255,7 +255,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
             automargin=True,
             tickmode="array",
             tickvals=[-1.5] + list(range(df_shap.shape[0])) + [df_shap.shape[0] + 0.5],
-            ticktext=["Chrono Age"] + df_shap.index.to_list() + ["EpImAge"],
+            ticktext=["Chrono Age"] + df_shap.index.to_list() + ["EpInflammAge"],
             tickfont=dict(size=16),
         )
         fig.update_xaxes(
@@ -321,7 +321,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
         fig.update_layout(barmode="relative")
         fig.update_layout(
             legend=dict(
-                title=dict(text="Immunomarkers disribution<br>in samples with same age", side="top"),
+                title=dict(text="Inflammatory Markers disribution<br>in samples with same age", side="top"),
                 orientation="h",
                 yanchor="bottom",
                 y=0.95,
@@ -388,26 +388,26 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
             data.loc[data.index, cpgs] = data_all.loc[data.index, cpgs]
                 
         # Models' inference
-        progress(0.9, desc="Immunology models' inference")
+        progress(0.9, desc="Inflammatory models' inference")
         for imm in imms:
             data[f"{imm}_log"] = models_imms[imm].predict(data)
-        progress(0.95, desc='EpImAge model inference')
-        data['EpImAge'] = model_age.predict(data.loc[:, [f"{imm}_log" for imm in imms]])
-        data['Age Acceleration'] = data['EpImAge'] - data['Age']
+        progress(0.95, desc='EpInflammAge model inference')
+        data['EpInflammAge'] = model_age.predict(data.loc[:, [f"{imm}_log" for imm in imms]])
+        data['Age Acceleration'] = data['EpInflammAge'] - data['Age']
         data.to_pickle(f'{dir_out}/{str(request.session_hash)}/data.pkl')
         
-        data_res = data[['Age', 'EpImAge', 'Age Acceleration'] + list(imms_log)]
+        data_res = data[['Age', 'EpInflammAge', 'Age Acceleration'] + list(imms_log)]
         data_res.rename(columns={f"{imm}_log": imm for imm in imms}).to_excel(f'{dir_out}/{str(request.session_hash)}/Result.xlsx', index_label='ID')
 
         if len(data_res) > 1:
-            mae = mean_absolute_error(data['Age'].values, data['EpImAge'].values)
-            rho = scipy.stats.pearsonr(data['Age'].values, data['EpImAge'].values).statistic
+            mae = mean_absolute_error(data['Age'].values, data['EpInflammAge'].values)
+            rho = scipy.stats.pearsonr(data['Age'].values, data['EpInflammAge'].values).statistic
         
         # Plot scatter
         progress(0.98, desc='Plotting scatter')
         fig = make_subplots(rows=1, cols=2, shared_yaxes=False, shared_xaxes=False, column_widths=[5, 3], horizontal_spacing=0.15)
-        min_plot_age = data[["Age", "EpImAge"]].min().min()
-        max_plot_age = data[["Age", "EpImAge"]].max().max()
+        min_plot_age = data[["Age", "EpInflammAge"]].min().min()
+        max_plot_age = data[["Age", "EpInflammAge"]].max().max()
         shift_plot_age = max_plot_age - min_plot_age
         min_plot_age -= 0.1 * shift_plot_age
         max_plot_age += 0.1 * shift_plot_age
@@ -426,7 +426,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
             go.Scatter(
                 name='Scatter',
                 x=data.loc[:, 'Age'].values,
-                y=data.loc[:, 'EpImAge'].values,
+                y=data.loc[:, 'EpInflammAge'].values,
                 text=data.index.values,
                 hovertext=data.index.values,
                 showlegend=False,
@@ -476,7 +476,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title='EpImAge', js=js_func, delete_cache
             row=1,
             col=1,
             automargin=True,
-            title_text=f"EpImAge",
+            title_text=f"EpInflammAge",
             # scaleanchor="x",
             # scaleratio=1,
             autorange=False,

configs/{immuno-regression → cytokines-regression}/DataConfig.yaml

@@ -1,14 +1,14 @@
-target: []
-continuous_cols:
-  - 'to_replace'
-categorical_cols: []
-date_columns: []
-encode_date_columns: true
-validation_split: 0.25
-continuous_feature_transform: yeo-johnson
-normalize_continuous_features: true
-quantile_noise: 0
-num_workers: 0
-pin_memory: true
-handle_unknown_categories: true
+target: []
+continuous_cols:
+  - 'to_replace'
+categorical_cols: []
+date_columns: []
+encode_date_columns: true
+validation_split: 0.25
+continuous_feature_transform: yeo-johnson
+normalize_continuous_features: true
+quantile_noise: 0
+num_workers: 0
+pin_memory: true
+handle_unknown_categories: true
 handle_missing_values: true

configs/{immuno-regression → cytokines-regression}/OptimizerConfig.yaml

@@ -1,23 +1,23 @@
-optimizer: Adam
-optimizer_params:
-  weight_decay: 1e-6
-
-lr_scheduler: ReduceLROnPlateau
-lr_scheduler_params:
-  mode: min
-  factor: 0.1
-  patience: 25
-  threshold: 1e-4
-
-#lr_scheduler: CosineAnnealingWarmRestarts
-#lr_scheduler_params:
-#  T_0: 10
-#  T_mult: 1
-#  eta_min: 1e-5
-
-#lr_scheduler: StepLR
-#lr_scheduler_params:
-#  step_size: 100
-#  gamma: 0.75
-
+optimizer: Adam
+optimizer_params:
+  weight_decay: 1e-6
+
+lr_scheduler: ReduceLROnPlateau
+lr_scheduler_params:
+  mode: min
+  factor: 0.1
+  patience: 25
+  threshold: 1e-4
+
+#lr_scheduler: CosineAnnealingWarmRestarts
+#lr_scheduler_params:
+#  T_0: 10
+#  T_mult: 1
+#  eta_min: 1e-5
+
+#lr_scheduler: StepLR
+#lr_scheduler_params:
+#  step_size: 100
+#  gamma: 0.75
+
 lr_scheduler_monitor_metric: valid_loss

configs/{immuno-regression → cytokines-regression}/TrainerConfig.yaml

@@ -1,26 +1,26 @@
-batch_size: 1024
-fast_dev_run: false
-max_epochs: 1000
-min_epochs: 1
-max_time: null
-accelerator: auto
-devices: -1
-accumulate_grad_batches: 1
-auto_lr_find: true
-check_val_every_n_epoch: 1
-gradient_clip_val: 0.0
-overfit_batches: 0.0
-profiler: null
-early_stopping: valid_loss
-early_stopping_min_delta: 0.000001
-early_stopping_mode: min
-early_stopping_patience: 50
-checkpoints: valid_loss
-checkpoints_path: ""
-checkpoints_every_n_epochs: 5
-checkpoints_mode: min
-checkpoints_save_top_k: 1
-load_best: true
-track_grad_norm: -1
-progress_bar: none
-seed: 1337
+batch_size: 1024
+fast_dev_run: false
+max_epochs: 1000
+min_epochs: 1
+max_time: null
+accelerator: auto
+devices: -1
+accumulate_grad_batches: 1
+auto_lr_find: true
+check_val_every_n_epoch: 1
+gradient_clip_val: 0.0
+overfit_batches: 0.0
+profiler: null
+early_stopping: valid_loss
+early_stopping_min_delta: 0.000001
+early_stopping_mode: min
+early_stopping_patience: 50
+checkpoints: valid_loss
+checkpoints_path: ""
+checkpoints_every_n_epochs: 5
+checkpoints_mode: min
+checkpoints_save_top_k: 1
+load_best: true
+track_grad_norm: -1
+progress_bar: none
+seed: 1337

configs/{immuno-regression → cytokines-regression}/models/CategoryEmbeddingModelConfig.yaml

@@ -1,26 +1,26 @@
-_module_src: models.category_embedding
-_model_name: CategoryEmbeddingModel
-_backbone_name: CategoryEmbeddingBackbone
-_config_name: CategoryEmbeddingModelConfig
-
-task: regression
-head: LinearHead
-head_config:
-  layers: ""
-  activation: ReLU
-  dropout: 0.1
-  use_batch_norm: false
-  initialization: xavier
-learning_rate: 1e-3
-loss: L1Loss
-metrics:
-  - mean_absolute_error
-  - pearson_corrcoef
-target_range: null
-seed: 1337
-
-layers: "256-128-64"
-activation: ReLU
-use_batch_norm: false
-initialization: xavier
-dropout: 0.1
+_module_src: models.category_embedding
+_model_name: CategoryEmbeddingModel
+_backbone_name: CategoryEmbeddingBackbone
+_config_name: CategoryEmbeddingModelConfig
+
+task: regression
+head: LinearHead
+head_config:
+  layers: ""
+  activation: ReLU
+  dropout: 0.1
+  use_batch_norm: false
+  initialization: xavier
+learning_rate: 1e-3
+loss: L1Loss
+metrics:
+  - mean_absolute_error
+  - pearson_corrcoef
+target_range: null
+seed: 1337
+
+layers: "256-128-64"
+activation: ReLU
+use_batch_norm: false
+initialization: xavier
+dropout: 0.1

configs/{immuno-regression → cytokines-regression}/models/DANetConfig.yaml

@@ -1,26 +1,26 @@
-_module_src: models.danet
-_model_name: DANetModel
-_backbone_name: DANetBackbone
-_config_name: DANetConfig
-
-task: regression
-head: LinearHead
-head_config:
-  layers: ""
-  activation: ReLU
-  dropout: 0.1
-  use_batch_norm: false
-  initialization: xavier
-learning_rate: 1e-3
-loss: L1Loss
-metrics:
-  - mean_absolute_error
-  - pearson_corrcoef
-target_range: null
-seed: 1337
-
-n_layers: 8
-abstlay_dim_1: 32
-abstlay_dim_2: null
-k: 5
-dropout_rate: 0.1
+_module_src: models.danet
+_model_name: DANetModel
+_backbone_name: DANetBackbone
+_config_name: DANetConfig
+
+task: regression
+head: LinearHead
+head_config:
+  layers: ""
+  activation: ReLU
+  dropout: 0.1
+  use_batch_norm: false
+  initialization: xavier
+learning_rate: 1e-3
+loss: L1Loss
+metrics:
+  - mean_absolute_error
+  - pearson_corrcoef
+target_range: null
+seed: 1337
+
+n_layers: 8
+abstlay_dim_1: 32
+abstlay_dim_2: null
+k: 5
+dropout_rate: 0.1

configs/{immuno-regression → cytokines-regression}/models/FTTransformerConfig.yaml

@@ -1,29 +1,29 @@
-_module_src: models.ft_transformer
-_model_name: FTTransformerModel
-_backbone_name: FTTransformerBackbone
-_config_name: FTTransformerConfig
-
-task: regression
-head: LinearHead
-head_config:
-  layers: ""
-  activation: ReLU
-  dropout: 0.1
-  use_batch_norm: false
-  initialization: xavier
-learning_rate: 1e-3
-loss: L1Loss
-metrics:
-  - mean_absolute_error
-  - pearson_corrcoef
-target_range: null
-seed: 1337
-
-attn_feature_importance: false
-num_heads: 8
-num_attn_blocks: 6
-attn_dropout: 0.1
-add_norm_dropout: 0.1
-ff_dropout: 0.1
-ff_hidden_multiplier: 4
+_module_src: models.ft_transformer
+_model_name: FTTransformerModel
+_backbone_name: FTTransformerBackbone
+_config_name: FTTransformerConfig
+
+task: regression
+head: LinearHead
+head_config:
+  layers: ""
+  activation: ReLU
+  dropout: 0.1
+  use_batch_norm: false
+  initialization: xavier
+learning_rate: 1e-3
+loss: L1Loss
+metrics:
+  - mean_absolute_error
+  - pearson_corrcoef
+target_range: null
+seed: 1337
+
+attn_feature_importance: false
+num_heads: 8
+num_attn_blocks: 6
+attn_dropout: 0.1
+add_norm_dropout: 0.1
+ff_dropout: 0.1
+ff_hidden_multiplier: 4
 transformer_activation: GEGLU

configs/{immuno-regression → cytokines-regression}/models/GANDALFConfig.yaml

@@ -1,24 +1,24 @@
-_module_src: models.gandalf
-_model_name: GANDALFModel
-_backbone_name: GANDALFBackbone
-_config_name: GANDALFConfig
-
-task: regression
-head: LinearHead
-head_config:
-  layers: ""
-  activation: ReLU
-  dropout: 0.1
-  use_batch_norm: false
-  initialization: xavier
-learning_rate: 1e-3
-loss: L1Loss
-metrics:
-  - mean_absolute_error
-  - pearson_corrcoef
-target_range: null
-seed: 1337
-
-gflu_stages: 6
-gflu_dropout: 0.0
-gflu_feature_init_sparsity: 0.3
+_module_src: models.gandalf
+_model_name: GANDALFModel
+_backbone_name: GANDALFBackbone
+_config_name: GANDALFConfig
+
+task: regression
+head: LinearHead
+head_config:
+  layers: ""
+  activation: ReLU
+  dropout: 0.1
+  use_batch_norm: false
+  initialization: xavier
+learning_rate: 1e-3
+loss: L1Loss
+metrics:
+  - mean_absolute_error
+  - pearson_corrcoef
+target_range: null
+seed: 1337
+
+gflu_stages: 6
+gflu_dropout: 0.0
+gflu_feature_init_sparsity: 0.3

configs/{immuno-regression → cytokines-regression}/models/TabNetModelConfig.yaml

@@ -1,29 +1,29 @@
-_module_src: models.tabnet
-_model_name: TabNetModel
-_backbone_name: TabNetBackbone
-_config_name: TabNetModelConfig
-
-task: regression
-head: LinearHead
-head_config:
-  layers: ""
-  activation: ReLU
-  dropout: 0.1
-  use_batch_norm: false
-  initialization: xavier
-learning_rate: 1e-3
-loss: L1Loss
-metrics:
-  - mean_absolute_error
-  - pearson_corrcoef
-target_range: null
-seed: 1337
-
-n_d: 8
-n_a: 8
-n_steps: 3
-gamma: 1.3
-n_independent: 2
-n_shared: 2
-virtual_batch_size: 512
+_module_src: models.tabnet
+_model_name: TabNetModel
+_backbone_name: TabNetBackbone
+_config_name: TabNetModelConfig
+
+task: regression
+head: LinearHead
+head_config:
+  layers: ""
+  activation: ReLU
+  dropout: 0.1
+  use_batch_norm: false
+  initialization: xavier
+learning_rate: 1e-3
+loss: L1Loss
+metrics:
+  - mean_absolute_error
+  - pearson_corrcoef
+target_range: null
+seed: 1337
+
+n_d: 8
+n_a: 8
+n_steps: 3
+gamma: 1.3
+n_independent: 2
+n_shared: 2
+virtual_batch_size: 512
 mask_type: sparsemax

models/{EpImAge → EpInflammAge}/config.yml

@@ -1,118 +1,118 @@
-target:
-- Age
-continuous_cols:
-- CXCL9_log
-- CCL11_log
-- IL27_log
-- IL5_log
-- CSF1_log
-- CCL2_log
-- IL1B_log
-- IL6_log
-- GCSF_log
-- CXCL10_log
-- VEGFA_log
-- TNF_log
-- PDGFB_log
-- IL8_log
-- PDGFA_log
-- IL12Bp40_log
-- IL15_log
-- CXCL1_log
-- CCL4_log
-- IFNA2_log
-- IL13_log
-- FLT3L_log
-- CD40LG_log
-- CCL22_log
-categorical_cols: []
-date_columns: []
-encode_date_columns: true
-validation_split: 0.25
-continuous_feature_transform: yeo-johnson
-normalize_continuous_features: true
-quantile_noise: 0
-num_workers: 0
-pin_memory: true
-handle_unknown_categories: true
-handle_missing_values: true
-task: regression
-head: LinearHead
-head_config:
-  layers: ''
-  activation: ReLU
-  dropout: 0.23377289048411876
-  use_batch_norm: false
-  initialization: xavier
-embedding_dims: null
-embedding_dropout: 0.0
-batch_norm_continuous_input: true
-learning_rate: 0.029204372085555506
-loss: L1Loss
-metrics:
-- mean_absolute_error
-- pearson_corrcoef
-metrics_prob_input:
-- false
-- false
-metrics_params:
-- {}
-- {}
-target_range: null
-virtual_batch_size: 256
-seed: 1984
-_module_src: models.danet
-_model_name: DANetModel
-_backbone_name: DANetBackbone
-_config_name: DANetConfig
-n_layers: 28
-abstlay_dim_1: 8
-abstlay_dim_2: 64
-k: 2
-dropout_rate: 0.2320482225690916
-block_activation: LeakyReLU
-batch_size: 1024
-data_aware_init_batch_size: 2000
-fast_dev_run: false
-max_epochs: 256
-min_epochs: 1
-max_time: null
-accelerator: auto
-devices: -1
-devices_list: null
-accumulate_grad_batches: 1
-auto_lr_find: false
-auto_select_gpus: true
-check_val_every_n_epoch: 1
-gradient_clip_val: 0.0
-overfit_batches: 0.0
-deterministic: false
-profiler: null
-early_stopping: valid_loss
-early_stopping_min_delta: 0.001
-early_stopping_mode: min
-early_stopping_patience: 20
-early_stopping_kwargs: {}
-checkpoints: valid_loss
-checkpoints_path: D:/YandexDisk/Work/bbd/immunology/003_EpImAge/imp_source(imm)_method(knn)_params(5)/no_harm/mrmr_100/EpImAge/pytorch_tabular
-checkpoints_every_n_epochs: 1
-checkpoints_name: null
-checkpoints_mode: min
-checkpoints_save_top_k: 1
-checkpoints_kwargs: {}
-load_best: true
-track_grad_norm: -1
-progress_bar: none
-precision: 32
-trainer_kwargs: {}
-optimizer: Adam
-optimizer_params:
-  weight_decay: 2.149173764897728e-08
-lr_scheduler: ReduceLROnPlateau
-lr_scheduler_params:
-  mode: min
-  factor: 0.7071915700100401
-  patience: 10
-  threshold: 0.0001
-lr_scheduler_monitor_metric: valid_loss
-enable_checkpointing: true
+target:
+- Age
+continuous_cols:
+- CXCL9_log
+- CCL11_log
+- IL27_log
+- IL5_log
+- CSF1_log
+- CCL2_log
+- IL1B_log
+- IL6_log
+- GCSF_log
+- CXCL10_log
+- VEGFA_log
+- TNF_log
+- PDGFB_log
+- IL8_log
+- PDGFA_log
+- IL12Bp40_log
+- IL15_log
+- CXCL1_log
+- CCL4_log
+- IFNA2_log
+- IL13_log
+- FLT3L_log
+- CD40LG_log
+- CCL22_log
+categorical_cols: []
+date_columns: []
+encode_date_columns: true
+validation_split: 0.25
+continuous_feature_transform: yeo-johnson
+normalize_continuous_features: true
+quantile_noise: 0
+num_workers: 0
+pin_memory: true
+handle_unknown_categories: true
+handle_missing_values: true
+task: regression
+head: LinearHead
+head_config:
+  layers: ''
+  activation: ReLU
+  dropout: 0.23377289048411876
+  use_batch_norm: false
+  initialization: xavier
+embedding_dims: null
+embedding_dropout: 0.0
+batch_norm_continuous_input: true
+learning_rate: 0.029204372085555506
+loss: L1Loss
+metrics:
+- mean_absolute_error
+- pearson_corrcoef
+metrics_prob_input:
+- false
+- false
+metrics_params:
+- {}
+- {}
+target_range: null
+virtual_batch_size: 256
+seed: 1984
+_module_src: models.danet
+_model_name: DANetModel
+_backbone_name: DANetBackbone
+_config_name: DANetConfig
+n_layers: 28
+abstlay_dim_1: 8
+abstlay_dim_2: 64
+k: 2
+dropout_rate: 0.2320482225690916
+block_activation: LeakyReLU
+batch_size: 1024
+data_aware_init_batch_size: 2000
+fast_dev_run: false
+max_epochs: 256
+min_epochs: 1
+max_time: null
+accelerator: auto
+devices: -1
+devices_list: null
+accumulate_grad_batches: 1
+auto_lr_find: false
+auto_select_gpus: true
+check_val_every_n_epoch: 1
+gradient_clip_val: 0.0
+overfit_batches: 0.0
+deterministic: false
+profiler: null
+early_stopping: valid_loss
+early_stopping_min_delta: 0.001
+early_stopping_mode: min
+early_stopping_patience: 20
+early_stopping_kwargs: {}
+checkpoints: valid_loss
+checkpoints_path: D:/YandexDisk/Work/bbd/immunology/003_EpImAge/imp_source(imm)_method(knn)_params(5)/no_harm/mrmr_100/EpImAge/pytorch_tabular
+checkpoints_every_n_epochs: 1
+checkpoints_name: null
+checkpoints_mode: min
+checkpoints_save_top_k: 1
+checkpoints_kwargs: {}
+load_best: true
+track_grad_norm: -1
+progress_bar: none
+precision: 32
+trainer_kwargs: {}
+optimizer: Adam
+optimizer_params:
+  weight_decay: 2.149173764897728e-08
+lr_scheduler: ReduceLROnPlateau
+lr_scheduler_params:
+  mode: min
+  factor: 0.7071915700100401
+  patience: 10
+  threshold: 0.0001
+lr_scheduler_monitor_metric: valid_loss
+enable_checkpointing: true

models/EpInflammAge/custom_params.sav

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e66e47b6f4047b2a3a2b3581ff778d82b6d9a193030ba322fd3a47baf67c0655
+size 127