BMList: A Comprehensive List of Popular Big Models
Captured source
source ↗BMList A Comprehensive List of Big Models | Medium
Sign up
Get app
Sign up
BMList: A Comprehensive List of Popular Big Models
4 min read
Dec 23, 2022
--
Share
Press enter or click to view image in full size
Do you know:
▶ Back in 2017, Google introduced a “big model” with 137 billion parameters? (Based on MoE and LSTM architecture)
▶It says that big models are difficult to train, but there are more than 100 released big models with more than 1 billion parameters?
▶ In addition to the common English and Chinese models, there are also big models for French, Korean, Russian, Arabic and other languages?
▶ In the era of big models, which companies and institutions have already won the race and currently possess the most experience and reserves for training big models?
▶ In the face of so many big models, we are already confused just to remember the names. When we need to know more about them, how do we choose and quickly find relevant resources?
Press enter or click to view image in full size
BMList is able to answer these questions!
OpenBMB has released BMList, a list of big models on GitHub. BMList sorts out the publicly available information about big models, comprehensively displays the model information through the Big Model List, and makes graph statistics of the model information in the Big Model Gallery. Two ways to get us a more comprehensive and intuitive understanding of the current big model information, check it out!
Big Model List
Big Model List displays information about big models in a multidimensional table. The list features the following:
▶ Comprehensive information
The list includes the release date, organization, language, number of parameters, and links to documentation, parameters, code, and API.
▶ Filter quickly
The list supports model filtering by tags such as organization, language, and domain, which facilitates users to quickly view the models related to a specific language, specific organization, or specific domain.
▶ Open source
Get OpenBMB’s stories in your inbox
Join Medium for free to get updates from this writer.
Subscribe
Subscribe
Remember me for faster sign in
The list organizes the parameters, code, and API open source links of big models, which facilitates users to quickly locate and experience the capabilities of big models.
Press enter or click to view image in full size
Big Model Gallery
Big Model Gallery provides statistical charts related to big model information. By publicly available information, we can observe trends in the big models and draw some interesting conclusions (the analysis is limited to the included data).
Press enter or click to view image in full size
As can be seen from this big model trend figure, English text models still dominate, and Chinese text models appear a little later than English text models, but there are corresponding Chinese text models in each order of magnitude. Also seen in the figure are Sparsely-Gate MoE, the earliest “grand model” in 2017, and M6–10T, the biggest of the current 10 trillion parametric models.
Press enter or click to view image in full size
In the chart of the number of models released by different organizations, we can see that Google has topped the pack with more than 20 big models, followed by Meta and Tsinghua University. In terms of the number of models, all the organizations are still roughly equal.
Press enter or click to view image in full size
As for the sum of parameters of all released models, we can find that among the top ten institutions, Alibaba, BAAI, Pengcheng Lab and Baidu rank first, third, ninth and tenth respectively, and Chinese institutions are slightly behind other institutions in terms of model parameters.
Participate in Contribution
BMList provides a very easy way to add new models. We agreed on two simple model selection rules:
▶ 01 Model parameters should be greater than 1 billion (i.e., greater than 1B parameters)
▶ 02 The model has been published publicly through a paper, blog, or press
Refer to the Contribution Guidelines: https://github.com/OpenBMB/BMList
All you need to do is submit a yaml file with the name of the big model to the big_models folder. After the PR is passed, the Big Model List and the Big Model gallery will be updated automatically.
In addition to adding new models, if you have new ideas and opinions on how to display data, model information fields, etc., you are also welcome to put forward in the community. In addition, since the information comes from open channels, the list may not be completely accurate in the content of the included models and model information. We also hope that you can actively offer valuable comments and suggestions through issues and PR!
Finally, we hope BMList can become a “memo” in the era of big models, so that we can witness the progress and development of the era of big models together!
Link to BMList
GitHub - OpenBMB/BMList: A List of Big Models
Welcome to BMList! We wish to use this list to show the recent trend of big models. In BMList, we list models that…
Written by OpenBMB
OpenBMB, known as Open Lab for Big Model Base, aims to build a large-scale pre-trained language model library and related tools
No responses yet
More from OpenBMB
·
Nov 17, 2022
OpenBMB: Big Models for Everyone
In recent years, as the pre-trained language model technology has triggered a performance revolution in artificial intelligence, maturity…
·
Nov 16, 2023
ModelBest Adopts the LLM-Based AI Agents to Launch ChatDev, its First SaaS Product
Breaking news! Here comes an AI-native app developed using the LLM-based AI agents.
·
Jul 15, 2023
Speak and draw! VisCPM：SOTA Open-source Chinese Multimodal Large Model
VisCPM is a series of open-source large multimodal models, supporting multimodal conversational and and text-to-image generation…
·
Jun 17, 2023
Official Tutorial — How to Fine-tune CPM-Bee on Basic Tasks
A formal and detailed tutorial of how to fine-tune the foundation model CPM-Bee and its data formats…
See all from OpenBMB
Recommended from Medium
In
by
Dr. Patricia Schmidt 🧠
·
Jan 14
As a Neuroscientist, I Quit These 5 Morning Habits That Destroy Your Brain
Most people do #1 within 10 minutes of waking (and it sabotages your entire day)
·
Jan 20
The Complete Guide to Fine-Tuning Embedding Models: From Theory to Production
A comprehensive guide to understanding,…
Excerpt shown — open the source for the full document.