ekurtic commited on
Commit
76e9580
1 Parent(s): 48e8dbd

Model release

Browse files
.gitattributes CHANGED
@@ -25,3 +25,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
25
  *.zip filter=lfs diff=lfs merge=lfs -text
26
  *.zstandard filter=lfs diff=lfs merge=lfs -text
27
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
25
  *.zip filter=lfs diff=lfs merge=lfs -text
26
  *.zstandard filter=lfs diff=lfs merge=lfs -text
27
  *tfevents* filter=lfs diff=lfs merge=lfs -text
28
+ *.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # oBERT-12-downstream-pruned-block4-90-QAT-squadv1
2
+
3
+ This model is obtained with [The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models](https://arxiv.org/abs/2203.07259).
4
+
5
+
6
+ It corresponds to the model presented in the `Table 3 - 12 Layers - Sparsity 90% - 4-block + QAT`.
7
+
8
+ ```
9
+ Pruning method: oBERT downstream block-4 + QAT
10
+ Paper: https://arxiv.org/abs/2203.07259
11
+ Dataset: SQuADv1
12
+ Sparsity: 90%
13
+ Number of layers: 12
14
+ ```
15
+
16
+ The dev-set performance of this model:
17
+ ```
18
+ EM = 78.84
19
+ F1 = 86.68
20
+ ```
21
+
22
+ Code: _coming soon_
23
+
24
+ ## BibTeX entry and citation info
25
+ ```bibtex
26
+ @article{kurtic2022optimal,
27
+ title={The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models},
28
+ author={Kurtic, Eldar and Campos, Daniel and Nguyen, Tuan and Frantar, Elias and Kurtz, Mark and Fineran, Benjamin and Goin, Michael and Alistarh, Dan},
29
+ journal={arXiv preprint arXiv:2203.07259},
30
+ year={2022}
31
+ }
32
+ ```
all_results.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d40d6856bcf734ec4e873ee2d5ebd1f12c4ec3ea4a21420b0650ee9654061a90
3
+ size 253
config.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba40a8b7c1cc13bc857bf31b02116d27d84738807807ff2d5a55c3ddf03b18c2
3
+ size 738
eval_nbest_predictions.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de32876da4d81a85309fe7f92de3ba19e7fe00ca66ae3af79397cdbd7b8259bf
3
+ size 49361811
eval_predictions.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a083bf3082b0ed9b1df6b71610550608b857a6d588481de5ad48d6c2a2c54685
3
+ size 589386
eval_results.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b66c85719f05c8793eac31d2968b97e5ed2a15cd6791fac28126c3e1d791787c
3
+ size 115
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb299f5c1d2bc10b5c53458a7ba3e3a44516d0cc5c146cdf3e1a3e10bee27481
3
+ size 436619715
recipe.yaml ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ !QuantizationModifier
2
+ disable_quantization_observer_epoch: 5
3
+ end_epoch: -1.0
4
+ freeze_bn_stats_epoch: 5
5
+ quantize_embeddings: 1
6
+ start_epoch: 0.0
7
+ submodules: ['qa_outputs', 'bert.embeddings', 'bert.encoder']
8
+
special_tokens_map.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:303df45a03609e4ead04bc3dc1536d0ab19b5358db685b6f3da123d05ec200e3
3
+ size 112
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5fd1c882abbd30517dced455a2c9768945ec726b96727927e4959348d9de550b
3
+ size 466081
tokenizer_config.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d3d4d2ced002f90895b722862032bf28a2b259a1d23151238381eecad83bb89
3
+ size 452
train_results.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9dc4f811bae9bae240cc6b6dc6f2d79aebe9521790222af6dce013923e98d0f
3
+ size 159
trainer_state.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a403352f671c1b4c676bb1b158090779ffd4ca545f1781f7c3145f76d523f9d4
3
+ size 7230
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80b318a85149e8f7a13a73b432b0ced47ea94cad20e278c502ef89b0d80cab5e
3
+ size 2607
vocab.txt ADDED
The diff for this file is too large to render. See raw diff