madlag commited on
Commit
6888c84
1 Parent(s): 3a7df1f

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -25,12 +25,12 @@ This model was created using the [nn_pruning](https://github.com/huggingface/nn_
25
 
26
  The model contains **42.0%** of the original weights **overall** (the embeddings account for a significant part of the model, and they are not pruned by this method).
27
 
28
- With a simple resizing of the linear matrices it ran **2.44x as fast as /home/lagunas/devel/hf/nn_pruning/nn_pruning/analysis/tmp_finetune** on the evaluation.
29
  This is possible because the pruning method lead to structured matrices: to visualize them, hover below on the plot to see the non-zero/zero parts of each matrix.
30
 
31
  <div class="graph"><script src="/madlag/bert-base-uncased-squadv1-x2.44-f87.7-d26-hybrid-filled-v1/raw/main/model_card/density_info.js" id="d5d1b3e9-73f5-4cfc-8e33-3745054bc7d0"></script></div>
32
 
33
- In terms of accuracy, its **F1 is 87.71**, compared with 88.5 for /home/lagunas/devel/hf/nn_pruning/nn_pruning/analysis/tmp_finetune, a **F1 drop of 0.79**.
34
 
35
  ## Fine-Pruning details
36
  This model was fine-tuned from the HuggingFace [model](https://huggingface.co//home/lagunas/devel/hf/nn_pruning/nn_pruning/analysis/tmp_finetune) checkpoint on [SQuAD1.1](https://rajpurkar.github.io/SQuAD-explorer), and distilled from the model [csarron/bert-base-uncased-squad-v1](https://huggingface.co/csarron/bert-base-uncased-squad-v1)
25
 
26
  The model contains **42.0%** of the original weights **overall** (the embeddings account for a significant part of the model, and they are not pruned by this method).
27
 
28
+ With a simple resizing of the linear matrices it ran **2.44x as fast as the original model** on the evaluation.
29
  This is possible because the pruning method lead to structured matrices: to visualize them, hover below on the plot to see the non-zero/zero parts of each matrix.
30
 
31
  <div class="graph"><script src="/madlag/bert-base-uncased-squadv1-x2.44-f87.7-d26-hybrid-filled-v1/raw/main/model_card/density_info.js" id="d5d1b3e9-73f5-4cfc-8e33-3745054bc7d0"></script></div>
32
 
33
+ In terms of accuracy, its **F1 is 87.71**, compared with 88.5 for the original model, a **F1 drop of 0.79**.
34
 
35
  ## Fine-Pruning details
36
  This model was fine-tuned from the HuggingFace [model](https://huggingface.co//home/lagunas/devel/hf/nn_pruning/nn_pruning/analysis/tmp_finetune) checkpoint on [SQuAD1.1](https://rajpurkar.github.io/SQuAD-explorer), and distilled from the model [csarron/bert-base-uncased-squad-v1](https://huggingface.co/csarron/bert-base-uncased-squad-v1)