cleaned up comments

src/models/cls_module.py

@@ -5,7 +5,6 @@
 # LICENSE file in the root directory of this source tree.
 
 
-
 import torch
 from lightning import LightningModule
 from torchmetrics import MinMetric, MaxMetric, MeanMetric
@@ -17,7 +16,6 @@
 from src.utils.metrics import MeanMetricWithCount
 
 
-
 class CLSLitModule(LightningModule):
     """Example of LightningModule for MNIST classification.
 
@@ -85,13 +83,6 @@ def __init__(
         self.dataset = None
         self.tuning = False
 
-        self.indices_to_check = [26291, 31336, 36921, 41564, 49741]
-        self.idx0_train_loss = MeanMetric()
-        self.idx1_train_loss = MeanMetric()
-        self.idx2_train_loss = MeanMetric()
-        self.idx3_train_loss = MeanMetric()
-        self.idx4_train_loss = MeanMetric()
-
     def on_train_start(self):
         # by default lightning executes validation step sanity checks before training starts,
         # so it's worth to make sure validation metrics don't store results from these checks
@@ -214,7 +205,6 @@ def configure_optimizers(self):
             https://lightning.ai/docs/pytorch/latest/common/lightning_module.html#configure-optimizers
         """
         if self.tuning:
-            #assert self.dataset is not None
             train_parameters = []
             for name, params in self.net.named_parameters():
                 if name.startswith('fc.'):

src/models/components/criterion.py

@@ -184,7 +184,6 @@ def forward(
         losses = self.common_step(output_dict, input_dict)
         if self.training:
             adjusted_loss = (torch.mean(losses) - self.flood_level).abs() + self.flood_level
-            #import IPython; IPython.embed()
         else:
             adjusted_loss = torch.mean(losses)
         return {constants.LOSS: adjusted_loss, constants.LOSSES: losses}
@@ -241,7 +240,6 @@ def forward(
 
         losses = self.common_step(output_dict, input_dict)
         if self.training:
-            #probs = torch.softmax(output_dict[constants.AUX_LOGITS], dim=1)
             aux_eval_losses = self.aux_step(
                 output_dict, input_dict, 'aux')
 

src/models/tpp/prob_dists.py

@@ -74,11 +74,6 @@ def forward(self, histories, log_times, masks):
         masks_without_last[torch.arange(batch_size), last_event_idx] = 0
         event_ll = (log_probs * masks_without_last).sum((1, 2)) - log_times.sum((1,2)) # (B,)
 
-        #if constants.LOG_MEAN in shared_data and constants.LOG_STD in shared_data:
-        #    log_mean = shared_data[constants.LOG_MEAN]
-        #    log_std = shared_data[constants.LOG_STD]
-        #else:
-        #    log_mean, log_std = 0.0, 1.0
         log_mean, log_std = 0.0, 1.0
 
         # compute predictions

src/models/tpp/thp/models.py

@@ -6,10 +6,8 @@
 
 
 import math
-import numpy as np
 import torch
 import torch.nn as nn
-import torch.nn.functional as F
 
 from src import constants
 from src.models.tpp.thp.layers import EncoderLayer, CrossAttnLayer
@@ -153,13 +151,6 @@ def __init__(
         nn.init.xavier_uniform_(self.query_proj.weight)
         nn.init.xavier_uniform_(self.key_proj.weight)
 
-        #self.attention = ScaledDotProductAttention(
-        #    temperature=d_k ** 0.5, attn_dropout=dropout)
-
-        #self.layer_norm = nn.LayerNorm(d_model, eps=1e-6)
-        #self.dropout = nn.Dropout(dropout)
-
-
     def temporal_enc(self, time, non_pad_mask):
         """
         Input: batch*seq_len.

src/models/tpp/tpp_aux_network.py

@@ -125,7 +125,6 @@ def __init__(self, name, activation, num_classes, d_model=256, d_inner=1024,
             num_latent=num_latent, vi_method=vi_method, num_z_samples=num_z_samples, compute_acc=compute_acc)
 
         aux_loss_path = aux_logit_path.replace('_logits', '_losses')
-        #aux_pred_path = aux_logit_path.replace('_logits', '_preds')
         aux_mu_path = aux_logit_path.replace('_logits', '_mus')
         aux_sigma_path = aux_logit_path.replace('_logits', '_sigmas')
         aux_log_weight_path = aux_logit_path.replace('_logits', '_log_weights')
@@ -141,9 +140,6 @@ def __init__(self, name, activation, num_classes, d_model=256, d_inner=1024,
         with open(aux_loss_path, "rb") as f:
             self.loss_dict = pickle.load(f)
 
-        #with open(aux_pred_path, "rb") as f:
-        #    self.pred_dict = pickle.load(f)
-
         with open(aux_mu_path, "rb") as f:
             self.mu_dict = pickle.load(f)
 

src/models/tpp_module.py

@@ -155,7 +155,6 @@ def on_validation_epoch_end(self):
             self.log("val/rmse_with_nll_best", self.val_rmse_with_nll_best.compute(), sync_dist=True, prog_bar=True)
             self.log("val/acc_with_nll_best", self.val_acc_with_nll_best.compute(), sync_dist=True, prog_bar=True)
 
-        #if val_rmse < prev_val_rmse_best:
         if val_rmse == val_rmse_best:
             self.val_nll_with_rmse_best.reset()
             self.val_nll_with_rmse_best(val_nll)