model documentation

#3
by nazneen - opened
Files changed (1) hide show
  1. README.md +186 -0
README.md ADDED
@@ -0,0 +1,186 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+
3
+ tags:
4
+ - bert
5
+ metrics:
6
+ - CodeBLEU
7
+ ---
8
+ # Model Card for plbart-base
9
+
10
+
11
+ # Model Details
12
+
13
+ ## Model Description
14
+
15
+ The PLBART model was proposed in Unified Pre-training for Program Understanding and Generation
16
+
17
+ - **Developed by:** UCLA NLP
18
+ - **Shared by [Optional]:** [Gunjan Chhablani](https://huggingface.co/gchhablani)
19
+ - **Model type:** Text2Text Generation
20
+ - **Language(s) (NLP):** More information needed
21
+ - **License:** More information needed
22
+ - **Related Models:** bert-base-multilingual-uncased
23
+ - **Parent Model:** plbart
24
+ - **Resources for more information:**
25
+ - [Associated Paper](https://arxiv.org/abs/2103.06333)
26
+ - [Model Documentation](https://huggingface.co/docs/transformers/model_doc/plbart)
27
+
28
+ # Uses
29
+
30
+ ## Direct Use
31
+
32
+ The pre-trained model plbart-base has been trained using multilingual denoising task
33
+
34
+ ## Downstream Use [Optional]
35
+
36
+ More information needed
37
+
38
+ ## Out-of-Scope Use
39
+ More information needed
40
+
41
+ # Bias, Risks, and Limitations
42
+
43
+
44
+ Significant research has explored bias and fairness issues with language models (see, e.g., [Sheng et al. (2021)](https://aclanthology.org/2021.acl-long.330.pdf) and [Bender et al. (2021)](https://dl.acm.org/doi/pdf/10.1145/3442188.3445922)). Predictions generated by the model may include disturbing and harmful stereotypes across protected classes; identity characteristics; and sensitive, social, and occupational groups.
45
+
46
+
47
+ ## Recommendations
48
+
49
+
50
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
51
+
52
+
53
+ # Training Details
54
+
55
+ ## Training Data
56
+
57
+ More information needed
58
+
59
+ ## Training Procedure
60
+
61
+
62
+
63
+ ### Preprocessing
64
+
65
+ The model creators note in the [associated paper](https://arxiv.org/pdf/2103.06333.pdf)
66
+ > We tokenize all the data with a sentencepiece model (Kudo and Richardson, 2018) learned on 1/5’th of the pre-training data. We train sentencepiece to learn 50,000 subword tokens. One key challenge to aggregate data from different modalities is that some modalities may have more data, such as we have 14 times more data in PL than NL. Therefore, we mix and up/down sample the data following Conneau and Lample (2019) to alleviate the bias towards PL.
67
+
68
+ ### Speeds, Sizes, Times
69
+ The model creators note in the [associated paper]()
70
+ > The effective batch size is maintained at 2048 instances.
71
+
72
+ # Evaluation
73
+
74
+
75
+ ## Testing Data, Factors & Metrics
76
+
77
+ ### Testing Data
78
+ The model creators note in the [associated paper](https://arxiv.org/pdf/2103.06333.pdf)
79
+ >CodeXGLUE (Lu et al., 2021) provided public dataset and corresponding train validation-test splits for all the tasks
80
+
81
+ ### Factors
82
+
83
+ More information needed
84
+
85
+ ### Metrics
86
+
87
+ More information needed
88
+
89
+ ## Results
90
+
91
+ More information needed
92
+
93
+ # Model Examination
94
+ More information needed
95
+
96
+ # Environmental Impact
97
+
98
+
99
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
100
+
101
+ - **Hardware Type:** 8 Nvidia GeForce RTX 2080 Ti GPUs
102
+ - **Hours used:** More information needed
103
+ - **Cloud Provider:** More information needed
104
+ - **Compute Region:** More information needed
105
+ - **Carbon Emitted:** More information needed
106
+
107
+ # Technical Specifications [optional]
108
+
109
+ ## Model Architecture and Objective
110
+
111
+ PLBart is a multilingual encoder-decoder (sequence-to-sequence) model primarily intended for code-to-text, text-to-code, code-to-code tasks. As the model is multilingual it expects the sequences in a different format. A special language id token is added in both the source and target text. The source text format is X [eos, src_lang_code] where X is the source text.
112
+
113
+ ## Compute Infrastructure
114
+ The model creators note in the [associated paper](https://arxiv.org/pdf/2103.06333.pdf)
115
+ > PLBART uses the same architecture as BARTbase (Lewis et al., 2020), it uses the sequence-to-sequence Transformer architecture (Vaswani et al., 2017), with 6 layers of encoder and 6 layers of decoder with model dimension of 768 and 12 heads (∼140M parameters). The only exception is, we include an additional layer normalization layer on top of both the encoder and decoder following Liu et al. (2020),
116
+
117
+ ### Hardware
118
+
119
+ More information needed
120
+
121
+ ### Software
122
+
123
+ More information needed
124
+
125
+ # Citation
126
+
127
+
128
+
129
+ **BibTeX:**
130
+ ```
131
+ @misc{https://doi.org/10.48550/arxiv.2103.06333,
132
+ doi = {10.48550/ARXIV.2103.06333},
133
+
134
+ url = {https://arxiv.org/abs/2103.06333},
135
+
136
+ author = {Ahmad, Wasi Uddin and Chakraborty, Saikat and Ray, Baishakhi and Chang, Kai-Wei},
137
+
138
+ keywords = {Computation and Language (cs.CL), Programming Languages (cs.PL), FOS: Computer and information sciences, FOS: Computer and information sciences},
139
+
140
+ title = {Unified Pre-training for Program Understanding and Generation},
141
+
142
+ publisher = {arXiv},
143
+
144
+ year = {2021},
145
+
146
+ copyright = {arXiv.org perpetual, non-exclusive license}
147
+ }
148
+
149
+
150
+ ```
151
+
152
+ **APA:**
153
+ More information needed
154
+
155
+ # Glossary [optional]
156
+
157
+ >CodeBLEU is a metric for measuring the quality of the synthesized code (Ren et al., 2020). Unlike BLEU, CodeBLEU also considers grammatical and logical correctness based on the abstract syntax tree and the data-flow structure.
158
+
159
+ # More Information [optional]
160
+
161
+ More information needed
162
+
163
+ # Model Card Authors [optional]
164
+
165
+
166
+ UCLA NLP in collaboration with Ezi Ozoani and the Hugging Face team
167
+
168
+ # Model Card Contact
169
+ More information needed
170
+
171
+ # How to Get Started with the Model
172
+
173
+ Use the code below to get started with the model.
174
+
175
+ <details>
176
+ <summary> Click to expand </summary>
177
+
178
+ ```python
179
+ from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
180
+
181
+ tokenizer = AutoTokenizer.from_pretrained("uclanlp/plbart-base")
182
+
183
+ model = AutoModelForSeq2SeqLM.from_pretrained("uclanlp/plbart-base")
184
+
185
+ ```
186
+ </details>