emre commited on
Commit
7f2e550
1 Parent(s): 5993880

create model_card

Browse files
Files changed (1) hide show
  1. README.md +59 -0
README.md ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - generated_from_trainer
5
+ - summarization
6
+ metrics:
7
+ - rouge
8
+ model-index:
9
+ - name: arxiv27k-t5-abst-title-gen/
10
+ results: []
11
+ ---
12
+
13
+ # arxiv27k-t5-abst-title-gen/
14
+
15
+ This model is a fine-tuned version of mt5-small on the arxiv-abstract-title dataset.
16
+ It achieves the following results on the evaluation set:
17
+ - Loss: 1.6002
18
+ - Rouge1: 32.8
19
+ - Rouge2: 21.9
20
+ - Rougel: 34.8
21
+ -
22
+
23
+ ## Model description
24
+
25
+ Model has been trained with a colab-pro notebook in 4 hours.
26
+
27
+ ## Intended uses & limitations
28
+
29
+ Can be used for generating journal titles from given abstracts
30
+
31
+ ### Training args
32
+ model_args = T5Args()
33
+ model_args.max_seq_length = 256
34
+ model_args.train_batch_size = 8
35
+ model_args.eval_batch_size = 8
36
+ model_args.num_train_epochs = 6
37
+ model_args.evaluate_during_training = False
38
+ model_args.use_multiprocessing = False
39
+ model_args.fp16 = False
40
+ model_args.save_steps = 40000
41
+ model_args.save_eval_checkpoints = False
42
+ model_args.save_model_every_epoch = True
43
+ model_args.output_dir = OUTPUT_DIR
44
+ model_args.no_cache = True
45
+ model_args.reprocess_input_data = True
46
+ model_args.overwrite_output_dir = True
47
+ model_args.num_return_sequences = 1
48
+
49
+
50
+ ### Framework versions
51
+
52
+ - Transformers 4.12.5
53
+ - Pytorch 1.10.0+cu111
54
+ - Datasets 1.15.1
55
+ - Tokenizers 0.10.3
56
+
57
+ ### Contact
58
+ detasar@gmail.com
59
+ Davut Emre Taşar