waidhoferj commited on
Commit
e748bc2
1 Parent(s): c8de10b

reduced requirements

Browse files
Files changed (3) hide show
  1. environment.yml +0 -10
  2. models/utils.py +1 -5
  3. requirements.txt +0 -1
environment.yml CHANGED
@@ -8,15 +8,11 @@ dependencies:
8
  - python=3.10
9
  - pytorch
10
  - torchaudio
11
- - torchvision
12
  - librosa
13
  - numpy
14
  - pandas
15
- - seaborn
16
- - matplotlib
17
  - bs4
18
  - requests
19
- - bidict
20
  - tqdm
21
  - pytorch-lightning
22
  - rich
@@ -25,9 +21,3 @@ dependencies:
25
  - transformers
26
  - accelerate
27
  - pytest
28
-
29
- - pip:
30
- - evaluate
31
- - wakepy
32
- - soundfile
33
- - youtube_dl
 
8
  - python=3.10
9
  - pytorch
10
  - torchaudio
 
11
  - librosa
12
  - numpy
13
  - pandas
 
 
14
  - bs4
15
  - requests
 
16
  - tqdm
17
  - pytorch-lightning
18
  - rich
 
21
  - transformers
22
  - accelerate
23
  - pytest
 
 
 
 
 
 
models/utils.py CHANGED
@@ -1,13 +1,9 @@
1
  import torch.nn as nn
2
  import torch
3
  import numpy as np
4
- import evaluate
5
  from sklearn.metrics import precision_score, recall_score, f1_score, accuracy_score
6
 
7
 
8
- accuracy = evaluate.load("accuracy")
9
-
10
-
11
  class LabelWeightedBCELoss(nn.Module):
12
  """
13
  Binary Cross Entropy loss that assumes each float in the final dimension is a binary probability distribution.
@@ -86,4 +82,4 @@ def get_id_label_mapping(labels: list[str]) -> tuple[dict, dict]:
86
 
87
  def compute_hf_metrics(eval_pred):
88
  predictions = np.argmax(eval_pred.predictions, axis=1)
89
- return accuracy.compute(predictions=predictions, references=eval_pred.label_ids)
 
1
  import torch.nn as nn
2
  import torch
3
  import numpy as np
 
4
  from sklearn.metrics import precision_score, recall_score, f1_score, accuracy_score
5
 
6
 
 
 
 
7
  class LabelWeightedBCELoss(nn.Module):
8
  """
9
  Binary Cross Entropy loss that assumes each float in the final dimension is a binary probability distribution.
 
82
 
83
  def compute_hf_metrics(eval_pred):
84
  predictions = np.argmax(eval_pred.predictions, axis=1)
85
+ return accuracy_score(y_true=eval_pred.label_ids, y_pred=predictions)
requirements.txt CHANGED
@@ -1,5 +1,4 @@
1
  torch
2
- torchvision
3
  torchaudio
4
  pytorch-lightning
5
  numpy
 
1
  torch
 
2
  torchaudio
3
  pytorch-lightning
4
  numpy