Adding ONNX file of this model

#2
by pszemraj - opened
Files changed (2) hide show
  1. README.md +20 -37
  2. config.json +20 -0
README.md CHANGED
@@ -1,4 +1,6 @@
1
  ---
 
 
2
  license:
3
  - cc-by-nc-sa-4.0
4
  - apache-2.0
@@ -55,54 +57,24 @@ widget:
55
  interested and anyone basical e may be applyind reaching the browing
56
  approach were
57
  - medical course audio transcription
58
- inference: False
59
- pipeline_tag: text2text-generation
 
 
 
 
60
  language:
61
  - en
 
62
  ---
63
 
64
 
65
  # bart-base-grammar-synthesis
66
 
67
- <a href="https://colab.research.google.com/gist/pszemraj/3e44e384dcd4614e1350d457bf9be8ad/bart-batch-grammar-check-correct-demo.ipynb">
68
- <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/>
69
- </a>
70
-
71
-
72
  This model is a fine-tuned version of [facebook/bart-base](https://huggingface.co/facebook/bart-base) on an expanded version of the JFLEG dataset.
73
 
74
  You can find other grammar-synthesis models by [searching for the grammar synthesis tag](https://huggingface.co/models?other=grammar%20synthesis)
75
 
76
- ## Basic Usage Example
77
-
78
- ### Installation
79
-
80
- First, make sure you have the `transformers` package installed. You can install it using pip:
81
-
82
- ```
83
- pip install -U transformers
84
- ```
85
-
86
- ### Usage
87
-
88
- ```python
89
- from transformers import pipeline
90
-
91
- # Initialize the text-generation pipeline for text correction
92
- corrector = pipeline("text2text-generation", "pszemraj/bart-base-grammar-synthesis")
93
-
94
- # Example text to correct
95
- raw_text = "The toweris 324 met (1,063 ft) tall, about height as .An 81-storey building, and biggest longest structure paris. Is square, measuring 125 metres (410 ft) on each side. During its constructiothe eiffel tower surpassed the washington monument to become the tallest man-made structure in the world, a title it held for 41 yearsuntilthe chryslerbuilding in new york city was finished in 1930. It was the first structure to goat a height of 300 metres. Due 2 the addition ofa brdcasting aerial at the t0pp of the twr in 1957, it now taller than chrysler building 5.2 metres (17 ft). Exxxcluding transmitters, eiffel tower is 2ndd tallest ree-standing structure in france after millau viaduct."
96
-
97
- # Correct the text using the text-generation pipeline
98
- corrected_text = corrector(raw_text)[0]["generated_text"]
99
-
100
- # Print the corrected text
101
- print(corrected_text)
102
- ```
103
-
104
- This example demonstrates how to use the text-generation pipeline to correct the grammar in a given text. The `corrector` pipeline is initialized with the "pszemraj/bart-base-grammar-synthesis" model, which is designed for grammar correction. The `corrector` pipeline takes the raw text as input and returns the corrected text. Make sure to install the required dependencies and models before running the code.
105
-
106
 
107
  ## Intended uses & limitations
108
 
@@ -129,3 +101,14 @@ The following hyperparameters were used during training:
129
  - lr_scheduler_type: cosine
130
  - lr_scheduler_warmup_ratio: 0.02
131
  - num_epochs: 3.0
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ languages:
3
+ - en
4
  license:
5
  - cc-by-nc-sa-4.0
6
  - apache-2.0
 
57
  interested and anyone basical e may be applyind reaching the browing
58
  approach were
59
  - medical course audio transcription
60
+ parameters:
61
+ max_new_tokens: 128
62
+ num_beams: 4
63
+ repetition_penalty: 1.21
64
+ length_penalty: 1
65
+ early_stopping: true
66
  language:
67
  - en
68
+ pipeline_tag: text2text-generation
69
  ---
70
 
71
 
72
  # bart-base-grammar-synthesis
73
 
 
 
 
 
 
74
  This model is a fine-tuned version of [facebook/bart-base](https://huggingface.co/facebook/bart-base) on an expanded version of the JFLEG dataset.
75
 
76
  You can find other grammar-synthesis models by [searching for the grammar synthesis tag](https://huggingface.co/models?other=grammar%20synthesis)
77
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
78
 
79
  ## Intended uses & limitations
80
 
 
101
  - lr_scheduler_type: cosine
102
  - lr_scheduler_warmup_ratio: 0.02
103
  - num_epochs: 3.0
104
+
105
+ ### Training results
106
+
107
+
108
+
109
+ ### Framework versions
110
+
111
+ - Transformers 4.28.1
112
+ - Pytorch 2.0.1+cu117
113
+ - Datasets 2.12.0
114
+ - Tokenizers 0.13.3
config.json CHANGED
@@ -48,6 +48,26 @@
48
  "num_hidden_layers": 6,
49
  "pad_token_id": 1,
50
  "scale_embedding": false,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
51
  "torch_dtype": "float32",
52
  "transformers_version": "4.28.1",
53
  "use_cache": true,
 
48
  "num_hidden_layers": 6,
49
  "pad_token_id": 1,
50
  "scale_embedding": false,
51
+ "task_specific_params": {
52
+ "summarization": {
53
+ "length_penalty": 1.0,
54
+ "max_length": 128,
55
+ "min_length": 12,
56
+ "num_beams": 4
57
+ },
58
+ "summarization_cnn": {
59
+ "length_penalty": 2.0,
60
+ "max_length": 142,
61
+ "min_length": 56,
62
+ "num_beams": 4
63
+ },
64
+ "summarization_xsum": {
65
+ "length_penalty": 1.0,
66
+ "max_length": 62,
67
+ "min_length": 11,
68
+ "num_beams": 6
69
+ }
70
+ },
71
  "torch_dtype": "float32",
72
  "transformers_version": "4.28.1",
73
  "use_cache": true,