Masioki commited on
Commit
d105ba9
1 Parent(s): df9391c

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +27 -11
README.md CHANGED
@@ -3,7 +3,25 @@ tags:
3
  - generated_from_trainer
4
  model-index:
5
  - name: prosody_gtsc_phi-3-mini-pitch
6
- results: []
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
  ---
8
 
9
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -11,21 +29,21 @@ should probably proofread and complete it, then remove this comment. -->
11
 
12
  # prosody_gtsc_phi-3-mini-pitch
13
 
14
- This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
15
 
16
- ## Model description
17
 
18
- More information needed
19
 
20
- ## Intended uses & limitations
 
 
 
21
 
22
- More information needed
23
 
24
  ## Training and evaluation data
25
 
26
- More information needed
27
-
28
- ## Training procedure
29
 
30
  ### Training hyperparameters
31
 
@@ -41,8 +59,6 @@ The following hyperparameters were used during training:
41
  - num_epochs: 20
42
  - mixed_precision_training: Native AMP
43
 
44
- ### Training results
45
-
46
 
47
 
48
  ### Framework versions
 
3
  - generated_from_trainer
4
  model-index:
5
  - name: prosody_gtsc_phi-3-mini-pitch
6
+ results:
7
+ - task:
8
+ type: dialogue act classification
9
+ dataset:
10
+ name: asapp/slue-phase-2
11
+ type: hvb
12
+ metrics:
13
+ - name: F1 macro E2E
14
+ type: F1 macro
15
+ value: 66.78
16
+ - name: F1 macro GT
17
+ type: F1 macro
18
+ value: 71.78
19
+ datasets:
20
+ - asapp/slue-phase-2
21
+ language:
22
+ - en
23
+ metrics:
24
+ - f1-macro
25
  ---
26
 
27
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
29
 
30
  # prosody_gtsc_phi-3-mini-pitch
31
 
 
32
 
33
+ Ground truth text with prosody encoding residual cross attention multi-label DAC
34
 
35
+ ## Model description
36
 
37
+ Prosody encoder: 2 layer transformer encoder with initial dense projection
38
+ Backbone: [Phi 3 mini](https://huggingface.co/microsoft/Phi-3-mini-4k-instruct)
39
+ Pooling: Self attention
40
+ Multi-label classification head: 2 dense layers with two dropouts 0.3 and Tanh activation inbetween
41
 
 
42
 
43
  ## Training and evaluation data
44
 
45
+ Trained on ground truth.
46
+ Evaluated on ground truth (GT) and normalized [Whisper small](https://huggingface.co/openai/whisper-small) transcripts (E2E).
 
47
 
48
  ### Training hyperparameters
49
 
 
59
  - num_epochs: 20
60
  - mixed_precision_training: Native AMP
61
 
 
 
62
 
63
 
64
  ### Framework versions