File size: 867 Bytes
e1b0c83
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4082075
e1b0c83
 
 
 
 
 
 
36f72e4
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
---
dataset_info:
  features:
  - name: prompt_text
    dtype: string
  - name: prompt_ids
    sequence: int64
  - name: decode_ids
    sequence: int64
  - name: prompt_pattern
    sequence:
      sequence: int64
  - name: decode_pattern
    sequence:
      sequence: int64
  - name: predictor_pattern
    sequence:
      sequence:
        sequence: int64
  splits:
  - name: train
    num_bytes: 297949158
    num_examples: 30000
  download_size: 30294841
  dataset_size: 297949158
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---


`decode_pattern.shape`: (num_layers, num_decoding_steps)
`predictor_pattern.shape`: (num_decoding_steps, num_layers, top3_indices)

We need to permute `predictor_pattern` via:

```python
predictor_pattern = predictor_pattern.permute(1,0,2) # (num_layers, num_decoding_steps, top3_indices)
```