Update README.md
Browse files
README.md
CHANGED
@@ -15,7 +15,7 @@ This model was created using the [nn_pruning](https://github.com/huggingface/nn_
|
|
15 |
|
16 |
The model contains **51%** of the original weights **overall** (the embeddings account for a significant part of the model, and they are not pruned by this method).
|
17 |
|
18 |
-
<div class="graph"><script src="/echarlaix/bert-base-uncased-sst2-acc91.1-d37-hybrid/raw/main/model_card/density_info.js" id="
|
19 |
|
20 |
In terms of perfomance, its **accuracy** is **91.17**.
|
21 |
|
@@ -25,7 +25,7 @@ This model is case-insensitive: it does not make a difference between english an
|
|
25 |
|
26 |
A side-effect of the block pruning method is that some of the attention heads are completely removed: 88 heads were removed on a total of 144 (61.1%).
|
27 |
Here is a detailed view on how the remaining heads are distributed in the network after pruning.
|
28 |
-
<div class="graph"><script src="/echarlaix/bert-base-uncased-sst2-acc91.1-d37-hybrid/raw/main/model_card/pruning_info.js" id="
|
29 |
|
30 |
## Details of the SST-2 dataset
|
31 |
|
15 |
|
16 |
The model contains **51%** of the original weights **overall** (the embeddings account for a significant part of the model, and they are not pruned by this method).
|
17 |
|
18 |
+
<div class="graph"><script src="/echarlaix/bert-base-uncased-sst2-acc91.1-d37-hybrid/raw/main/model_card/density_info.js" id="ca8aceb6-0975-4d02-b19f-4c1a5df8bf71"></script></div>
|
19 |
|
20 |
In terms of perfomance, its **accuracy** is **91.17**.
|
21 |
|
25 |
|
26 |
A side-effect of the block pruning method is that some of the attention heads are completely removed: 88 heads were removed on a total of 144 (61.1%).
|
27 |
Here is a detailed view on how the remaining heads are distributed in the network after pruning.
|
28 |
+
<div class="graph"><script src="/echarlaix/bert-base-uncased-sst2-acc91.1-d37-hybrid/raw/main/model_card/pruning_info.js" id="cfd2f3a7-70e9-40d6-824f-a6908fc889de"></script></div>
|
29 |
|
30 |
## Details of the SST-2 dataset
|
31 |
|