ArthurZ HF staff commited on
Commit
449ab1c
1 Parent(s): 6fb0883

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +110 -23
README.md CHANGED
@@ -1,47 +1,134 @@
1
  ---
 
 
2
  tags:
3
- - generated_from_keras_callback
4
- model-index:
5
- - name: opt-125m
6
- results: []
 
7
  ---
8
 
9
- <!-- This model card has been generated automatically according to the information Keras had access to. You should
10
- probably proofread and complete it, then remove this comment. -->
 
 
 
 
11
 
12
- # opt-125m
13
 
14
- This model was trained from scratch on an unknown dataset.
15
- It achieves the following results on the evaluation set:
16
 
17
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
18
  ## Model description
19
 
20
- More information needed
 
21
 
 
 
22
  ## Intended uses & limitations
23
 
24
- More information needed
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
25
 
26
- ## Training and evaluation data
27
 
28
- More information needed
 
 
29
 
30
  ## Training procedure
31
 
32
- ### Training hyperparameters
33
 
34
- The following hyperparameters were used during training:
35
- - optimizer: None
36
- - training_precision: float32
37
 
38
- ### Training results
39
 
 
 
40
 
 
41
 
42
- ### Framework versions
43
 
44
- - Transformers 4.20.0.dev0
45
- - TensorFlow 2.9.1
46
- - Datasets 2.2.2
47
- - Tokenizers 0.12.1
 
 
 
 
 
 
 
1
  ---
2
+ language: en
3
+ inference: false
4
  tags:
5
+ - text-generation
6
+ - opt
7
+
8
+ license: other
9
+ commercial: false
10
  ---
11
 
12
+ # OPT : Open Pre-trained Transformer Language Models
13
+
14
+ OPT was first introduced in [Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) and first released in [metaseq's repository](https://github.com/facebookresearch/metaseq) on May 3rd 2022 by Meta AI.
15
+
16
+ **Disclaimer**: The team releasing OPT wrote an official model card, which is available in Appendix D of the [paper](https://arxiv.org/pdf/2205.01068.pdf).
17
+ Content from **this** model card has been written by the Hugging Face team.
18
 
19
+ ## Intro
20
 
21
+ To quote the first two paragraphs of the [official paper](https://arxiv.org/abs/2205.01068)
 
22
 
23
 
24
+ > Large language models trained on massive text collections have shown surprising emergent
25
+ > capabilities to generate text and perform zero- and few-shot learning. While in some cases the public
26
+ > can interact with these models through paid APIs, full model access is currently limited to only a
27
+ > few highly resourced labs. This restricted access has limited researchers’ ability to study how and
28
+ > why these large language models work, hindering progress on improving known challenges in areas
29
+ > such as robustness, bias, and toxicity.
30
+
31
+ > We present Open Pretrained Transformers (OPT), a suite of decoder-only pre-trained transformers ranging from 125M
32
+ > to 175B parameters, which we aim to fully and responsibly share with interested researchers. We train the OPT models to roughly match
33
+ > the performance and sizes of the GPT-3 class of models, while also applying the latest best practices in data
34
+ > collection and efficient training. Our aim in developing this suite of OPT models is to enable reproducible and responsible research at scale, and
35
+ > to bring more voices to the table in studying the impact of these LLMs. Definitions of risk, harm, bias, and toxicity, etc., should be articulated by the
36
+ > collective research community as a whole, which is only possible when models are available for study.
37
+
38
  ## Model description
39
 
40
+ OPT was predominantly pretrained with English text, but a small amount of non-English data is still present within the training corpus via CommonCrawl. The model was pretrained using a causal language modeling (CLM) objective.
41
+ OPT belongs to the same family of decoder-only models like [GPT-3](https://arxiv.org/abs/2005.14165). As such, it was pretrained using the self-supervised causal language modedling objective.
42
 
43
+ For evaluation, OPT follows [GPT-3](https://arxiv.org/abs/2005.14165) by using their prompts and overall experimental setup. For more details, please read
44
+ the [official paper](https://arxiv.org/abs/2205.01068).
45
  ## Intended uses & limitations
46
 
47
+ The pretrained-only model can be used for prompting for evaluation of downstream tasks as well as text generation.
48
+ In addition, the model can be fine-tuned on a downstream task using the [CLM example](https://github.com/huggingface/transformers/tree/main/examples/pytorch/language-modeling). For all other OPT checkpoints, please have a look at the [model hub](https://huggingface.co/models?filter=opt).
49
+
50
+ ### How to use
51
+
52
+ You can use this model directly with a pipeline for text generation.
53
+
54
+ ```python
55
+ >>> from transformers import pipeline
56
+
57
+ >>> generator = pipeline('text-generation', model="facebook/opt-125m")
58
+ >>> generator("Hello, I'm am conscious and")
59
+ [{'generated_text': "Hello, I'm am conscious and conscious :) :) Anyway��極��極��極��極��極��極��極��極��極"}]
60
+ ```
61
+
62
+ By default, generation is deterministic. In order to use the top-k sampling, please set `do_sample` to `True`.
63
+
64
+ ```python
65
+ >>> from transformers import pipeline, set_seed
66
+
67
+ >>> set_seed(32)
68
+ >>> generator = pipeline('text-generation', model="facebook/opt-125m", do_sample=True)
69
+ >>> generator("Hello, I'm am conscious and")
70
+ [{'generated_text': "Hello, I'm am conscious and active observer!! HmmregorCLASSIFIEDドラゴン覚醒ドラゴンドラゴン覚醒覚醒ドラゴン"}]
71
+ ```
72
+
73
+ ### Limitations and bias
74
+
75
+ As mentioned in Meta AI's model card, given that the training data used for this model contains a lot of
76
+ unfiltered content from the internet, which is far from neutral the model is strongly biased :
77
+
78
+ > Like other large language models for which the diversity (or lack thereof) of training
79
+ > data induces downstream impact on the quality of our model, OPT-175B has limitations in terms
80
+ > of bias and safety. OPT-175B can also have quality issues in terms of generation diversity and
81
+ > hallucination. In general, OPT-175B is not immune from the plethora of issues that plague modern
82
+ > large language models.
83
+
84
+ This bias will also affect all fine-tuned versions of this model.
85
+
86
+ ## Training data
87
+
88
+ The Meta AI team wanted to train this model on a corpus as large as possible. It is composed of the union of the following 5 filtered datasets of textual documents:
89
+
90
+ - BookCorpus, which consists of more than 10K unpublished books,
91
+ - CC-Stories, which contains a subset of CommonCrawl data filtered to match the
92
+ story-like style of Winograd schemas,
93
+ - The Pile, from which * Pile-CC, OpenWebText2, USPTO, Project Gutenberg, OpenSubtitles, Wikipedia, DM Mathematics and HackerNews* were included.
94
+ - Pushshift.io Reddit dataset that was developed in Baumgartner et al. (2020) and processed in
95
+ Roller et al. (2021)
96
+ - CCNewsV2 containing an updated version of the English portion of the CommonCrawl News
97
+ dataset that was used in RoBERTa (Liu et al., 2019b)
98
+
99
+ The final training data contains 180B tokens corresponding to 800GB of data. The validation split was made of 200MB of the pretraining data, sampled proportionally
100
+ to each dataset’s size in the pretraining corpus.
101
+
102
+ The dataset might contains offensive content as parts of the dataset are a subset of
103
+ public Common Crawl data, along with a subset of public Reddit data, which could contain sentences
104
+ that, if viewed directly, can be insulting, threatening, or might otherwise cause anxiety.
105
 
106
+ ### Collection process
107
 
108
+ The dataset was collected form internet, and went through classic data processing algorithms and
109
+ re-formatting practices, including removing repetitive/non-informative text like *Chapter One* or
110
+ *This ebook by Project Gutenberg.*
111
 
112
  ## Training procedure
113
 
 
114
 
 
 
 
115
 
116
+ ### Preprocessing
117
 
118
+ The texts are tokenized using the **GPT2** byte-level version of Byte Pair Encoding (BPE) (for unicode characters) and a
119
+ vocabulary size of 50272. The inputs are sequences of 2048 consecutive tokens.
120
 
121
+ The 175B model was trained on 992 *80GB A100 GPUs*. The training duration was roughly ~33 days of continuous training.
122
 
123
+ ### BibTeX entry and citation info
124
 
125
+ ```bibtex
126
+ @misc{zhang2022opt,
127
+ title={OPT: Open Pre-trained Transformer Language Models},
128
+ author={Susan Zhang and Stephen Roller and Naman Goyal and Mikel Artetxe and Moya Chen and Shuohui Chen and Christopher Dewan and Mona Diab and Xian Li and Xi Victoria Lin and Todor Mihaylov and Myle Ott and Sam Shleifer and Kurt Shuster and Daniel Simig and Punit Singh Koura and Anjali Sridhar and Tianlu Wang and Luke Zettlemoyer},
129
+ year={2022},
130
+ eprint={2205.01068},
131
+ archivePrefix={arXiv},
132
+ primaryClass={cs.CL}
133
+ }
134
+ ```