ekurtic commited on
Commit
fc083b8
1 Parent(s): e9864f2

Initial commit

Browse files
README.md ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - bert
4
+ - mobilebert
5
+ - oBERT
6
+ language: en
7
+ datasets: squad
8
+ ---
9
+ # mobilebert-uncased-finetuned-squadv1
10
+
11
+ This model is a finetuned version of the [mobilebert-uncased](https://huggingface.co/google/mobilebert-uncased/tree/main) model on the SQuADv1 task.
12
+
13
+ It is produced as part of the work on the paper [The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models](https://arxiv.org/abs/2203.07259).
14
+
15
+ SQuADv1 dev-set:
16
+ ```
17
+ EM = 83.96
18
+ F1 = 90.90
19
+ ```
20
+
21
+ Code: [https://github.com/neuralmagic/sparseml/tree/main/research/optimal_BERT_surgeon_oBERT](https://github.com/neuralmagic/sparseml/tree/main/research/optimal_BERT_surgeon_oBERT)
22
+
23
+ If you find the model useful, please consider citing our work.
24
+
25
+ ## Citation info
26
+ ```bibtex
27
+ @article{kurtic2022optimal,
28
+ title={The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models},
29
+ author={Kurtic, Eldar and Campos, Daniel and Nguyen, Tuan and Frantar, Elias and Kurtz, Mark and Fineran, Benjamin and Goin, Michael and Alistarh, Dan},
30
+ journal={arXiv preprint arXiv:2203.07259},
31
+ year={2022}
32
+ }
33
+ ```
config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "neuralmagic/mobilebert-uncased-finetuned-squadv1",
3
+ "architectures": [
4
+ "MobileBertForQuestionAnswering"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.0,
7
+ "classifier_activation": false,
8
+ "classifier_dropout": null,
9
+ "embedding_size": 128,
10
+ "hidden_act": "relu",
11
+ "hidden_dropout_prob": 0.0,
12
+ "hidden_size": 512,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 512,
15
+ "intra_bottleneck_size": 128,
16
+ "key_query_shared_bottleneck": true,
17
+ "layer_norm_eps": 1e-12,
18
+ "max_position_embeddings": 512,
19
+ "model_type": "mobilebert",
20
+ "normalization_type": "no_norm",
21
+ "num_attention_heads": 4,
22
+ "num_feedforward_networks": 4,
23
+ "num_hidden_layers": 24,
24
+ "pad_token_id": 0,
25
+ "torch_dtype": "float32",
26
+ "transformers_version": "4.18.0.dev0",
27
+ "trigram_input": true,
28
+ "true_hidden_size": 128,
29
+ "type_vocab_size": 2,
30
+ "use_bottleneck": true,
31
+ "use_bottleneck_attention": false,
32
+ "vocab_size": 30522
33
+ }
eval_results.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "eval_exact_match": 83.96404919583728,
4
+ "eval_f1": 90.90359324656357,
5
+ "eval_samples": 10784
6
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:100120918cf6e8aea2737c46435f052f6a393dc035a58966818435967c1695d8
3
+ size 98801471
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": null, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "integrations/huggingface-transformers/output_dir/teachers/squad/bert_large_uncased_FP32", "tokenizer_class": "BertTokenizer"}
vocab.txt ADDED
The diff for this file is too large to render. See raw diff