Hugging Face's logo

ccdv
/
lsg-bart-base-4096-pubmed

Summarization
Transformers
PyTorch
English
bart
text2text-generation
custom_code
Model card Files Community
1
ccdv commited on
Commit
d36c8eb
•
1 Parent(s): aba9d66

readme

Browse files
Files changed (1) hide show
  1. README.md +2 -1
README.md CHANGED
@@ -55,13 +55,14 @@ With smaller block size (lower ressources):
55
 
56
  | Length | Sparse Type | Block Size | Sparsity | Connexions | R1 | R2 | RL | RLsum |
57
  |:------ |:------------ |:---------- |:-------- | :--------- |:----- |:----- |:----- |:----- |
 
 
58
  | 4096 | Pooling | 32 | 4 | 160 | 44.60 | 19.35 | 26.83 | 40.85 |
59
  | 4096 | Stride | 32 | 4 | 160 | 45.52 | 20.07 | 27.39 | 41.75 |
60
  | 4096 | Block Stride | 32 | 4 | 160 | 45.30 | 19.89 | 27.22 | 41.54 |
61
  | 4096 | Norm | 32 | 4 | 160 | 44.30 | 19.05 | 26.57 | 40.47 |
62
  | 4096 | LSH | 32 | 4 | 160 | 44.53 | 19.27 | 26.84 | 40.74 |
63
 
64
-
65
  ## Model description
66
  The model relies on Local-Sparse-Global attention to handle long sequences:
67
  ![attn](attn.png)
 
55
 
56
  | Length | Sparse Type | Block Size | Sparsity | Connexions | R1 | R2 | RL | RLsum |
57
  |:------ |:------------ |:---------- |:-------- | :--------- |:----- |:----- |:----- |:----- |
58
+ | 4096 | Local | 64 | 0 | 192 | 45.74 | 20.26 | 27.51 | 41.99 |
59
+ | 4096 | Local | 32 | 0 | 96 | 42.69 | 17.83 | 25.62 | 38.89 |
60
  | 4096 | Pooling | 32 | 4 | 160 | 44.60 | 19.35 | 26.83 | 40.85 |
61
  | 4096 | Stride | 32 | 4 | 160 | 45.52 | 20.07 | 27.39 | 41.75 |
62
  | 4096 | Block Stride | 32 | 4 | 160 | 45.30 | 19.89 | 27.22 | 41.54 |
63
  | 4096 | Norm | 32 | 4 | 160 | 44.30 | 19.05 | 26.57 | 40.47 |
64
  | 4096 | LSH | 32 | 4 | 160 | 44.53 | 19.27 | 26.84 | 40.74 |
65
 
 
66
  ## Model description
67
  The model relies on Local-Sparse-Global attention to handle long sequences:
68
  ![attn](attn.png)