From 71855c8d912d46fe3884d4593f28e59013df9239 Mon Sep 17 00:00:00 2001
From: umlauf <>
Date: Fri, 24 Feb 2023 13:43:55 +0100
Subject: [PATCH] Comments/Doc CEL
---
Code/train.py | 39 +++++++++++++++++++++------------------
1 file changed, 21 insertions(+), 18 deletions(-)
diff --git a/Code/train.py b/Code/train.py
index a0abfdb..e0ce706 100644
--- a/Code/train.py
+++ b/Code/train.py
@@ -139,34 +139,37 @@ def train(model, name, imdb, seed,mixup,lambda_value, mixepoch, tmix, mixlayer,
return evaluation_test, evaluation_train
-def cross_entropy(logits, target, l):
+def cross_entropy(logits, target):
+ """
+ Computes the cross-entropy loss between the predicted logits and the target labels.
+
+ Args:
+ - logits (torch.Tensor): A tensor of shape (batch_size, num_classes) representing the predicted logits for each input example.
+ - target (torch.Tensor): A tensor of shape (batch_size,) representing the target labels for each input example.
+
+ Returns:
+ - batch_loss (torch.Tensor): A scalar tensor representing the average cross-entropy loss across the batch.
+ """
results = torch.tensor([], device='cuda')
for i in range (logits.shape[0]):
- lg = logits[i:i+1,:] #comment to explain the process in this Code Line
+ lg = logits[i:i+1,:] # Extract a row from the logits tensor for the i-th input example in the batch
t = target[i]
- #makes the logits in log (base e) probabilities
- logprobs = torch.nn.functional.log_softmax(lg, dim=1)
- value = t.item() #gets Item (0. or 1.)
- if value == 1 or value == 0:
+ logprobs = torch.nn.functional.log_softmax(lg, dim=1) #logits in log (base e) probabilities
+ value = t.item() #get scalar value
+ if value == 1 or value == 0: #check if non-mixed label
one_hot = torch.tensor([1-value,value], device='cuda:0') #creating one-hot vector e.g. [0. ,1.]
- #class 1 and 2 mixed
loss_clear_labels = -((one_hot[0] * logprobs[0][0]) + (one_hot[1] * logprobs[0][1]))
+ #calculation with indexing (- 1-label * )
results = torch.cat((loss_clear_labels.view(1), results), dim=0)
else:
value_r = round(value, 1) #to make it equal to lambda_value e.g. 0.4
- #Wert mit Flag
- mixed_vec = torch.tensor([value_r, 1-value_r])
- print("Mixed Vec: ", mixed_vec)
- logprobs = torch.nn.functional.log_softmax(lg, dim=1)
- print("Log:", logprobs)
- #loss_mixed_labels = -torch.mul(mixed_vec, logprobs).sum()
+ mixed_vec = torch.tensor([value_r, 1-value_r]) #creating on-hot mixed vec.
+ logprobs = torch.nn.functional.log_softmax(lg, dim=1)#logits in log probabilities
loss_mixed_labels = -((mixed_vec[0] * logprobs[0][0]) + (mixed_vec[1] * logprobs[0][1]))
- print("Loss Mixed Lables l: ", loss_mixed_labels)
- results = torch.cat((loss_mixed_labels.view(1), results), dim=0)
- print("Results Mixed 1: ", results)
+ #calculation for mixed with indexing
+ results = torch.cat((loss_mixed_labels.view(1), results), dim=0)#append resultts to result tensor
print("ALL BATCH Results: ", results)
- batch_loss = results.mean() #compute average
- #print("Batch Loss: ", batch_loss)
+ batch_loss = results.mean() #compute average of all results
return batch_loss
--
GitLab