Hejing Community
Heywhale Community,Heywhale has over 400,000 data scientist and AI developer users
HeyWhale Community Official Website, with over 400,000 data scientists and AI developers
HeyWhale Community Overview
HeyWhale Community, with over 400,000 data scientists and AI developers, is one of the communities that stands out in the quality, quantity, and activity of talents in the AI and data science field. Stay updated with community dynamics, follow data science and AI development content, and focus on cutting-edge exploration.
HeyWhale Community Official Website:https://www.heywhale.com/
How is HeyWhale?
HeyWhale CommunityAn early domestic data analysis platform with clean and complete datasets, including retail, automotive, environmental, and mental health content.
The page layout is fresh and clear, with comprehensive data sets that allowreal-timeviewing of popular data sets and page views.
China's first large language model application competition, one-stop package of training + datasets + computing power | ChatGLM Practice Competition · Academic Application
HeyWhale Community Latest Competition
Since ChatGPT became popular, Xiao Xiaojing has come up with a big move! HeyWhale Technology has collaborated with many industry leaders and launchedChina's first large language model application competition，Training+Datasets+Computing Power All-in-Oneto provide an excellent practice platform for enthusiasts of large model research and development, and help developers explore the engineering and productization of large model technology!
Don't miss out on this opportunity to board the large model ship!
'ChatGLM Practice Competition · Academic Application'
· Event Background ·
Since the end of 2022, large language models have been widely applied in various industries, and many influential applications have emerged around academic tool development.
On the other hand, on March 14, 2023, Zhipu AI and Tsinghua University jointly released the 'ChatGLM-6B Open Model', which attracted over 1 million downloads and installations in less than a month and ranked first on the Hugging Face (HF) global large model download list for 12 consecutive days, making a significant impact in open-source communities at home and abroad.
To maximize the use of the ChatGLM-6B open-source model to promote the development of research tools, we have launched the'ChatGLM Practice Competition · Academic Application'in collaboration with AMiner, a domestically influential academic platform.
The central theme of this competition is how to use the ChatGLM-6B open-source model to promote the optimization of academic tools. At the same time, we also hope to provide a practice platform for enthusiasts who are willing to engage in large model research and development.We will provide participants with ChatGLM-6B
model fine-tuning training,relevant data,and possiblecomputing resources.
· Event Schedule ·
Registration&Development: Participants must register and submit project solutions on the HeyWhale competition platform from May 8 to June 13.Training: We will organize a technical training session on ChatGLM and the AMiner academic platform from May 8 to May 14.
Review: An expert review committee will evaluate the projects from June 17 to June 21.Awarding: Certificates, cash rewards, and computing power support will be awarded on June 24, and online exchange and sharing sessions will be organized for winners.
· Competition Description ·Scenario 1: Paper Reading
Track 1: Paper Discipline Classification (Easy)Task Description
: Accurately classify papers into 42 natural disciplines based on titles and abstracts, which may be single disciplines or interdisciplinary, with an accuracy of over 90%.
Track 2: Q&A Research Knowledge Base (Medium)
- Task Description
: Upload PDF papers to build a vectorized research knowledge base and perform free Q&A within the knowledge base, requiring relatively professional answers and attaching relevant files.Track 3: Paper Review and Comparative Analysis (Medium)
- Task Description
: Conduct a review or comparative analysis of the background, problems, methods, experiments, and conclusions of multiple papers given, including:– case1: Summarize the achievements of 5 papers;
- – case2: Analyze the differences in experimental methods between two papers.
Scenario 2: Submission and ReviewTrack 4: Journal/Conference Submission Recommendation (Medium)
Task Description
: Upload PDF papers to build a vectorized research knowledge base and perform free Q&A within the knowledge base, requiring relatively professional answers and attaching relevant files.
Track 5: Review Response (Medium)
- Task Description
: Fine-tune a review response model based on Openreview data.Scenario 3: Paper Discovery
- Track 6: Paper Retrieval (Hard)
Task Description: Perform separate and mixed retrieval of concepts, questions, and entities, such as:
1. Papers on the application of knowledge graphs (concepts) in product recommendation (problems);
- 2. Representative papers on large language models (concepts) by OpenAI and Google (entities);
3. How to solve search intent recognition (problems);4. What are the latest papers on the progress of ChatGPT (concepts)?
Track 7: Paper Recommendation and Technology Intelligence Generation (Hard)
Task Description
: Based on user profiles (domain keywords + search browsing behavior), use a large model to build an academic version of user profiles, and based on paper titles, abstracts, authors, and journals, build an academic version of paper profiles. Recommend 3-5 papers to users daily from the latest papers added, and extract summaries with intelligent attributes from the titles and abstracts of papers.
· Award Setup ·
- This event awards prizes based on the tasks registered for participation, with one best practice case awarded for each task, and awards set up according to the difficulty of the tasks as follows:
Hard Level:– 5,000 RMB GPU cloud computing resources;
– 5,000 RMB ChatGLM API support;
– ChatGLM official certification excellent certificate.
Medium Level:
– 3,000 RMB GPU cloud computing resources;
– 3,000 RMB ChatGLM API support;
– ChatGLM official certification excellent certificate.
Easy Level:
– 2,000 RMB GPU cloud computing resources;
– 2,000 RMB ChatGLM API support;
– ChatGLM official certification excellent certificate.
· Event Organization ·
Organizer
Zhipu AI
Co-organizer
Team Formation Deadline: 2023-06-13 (12:00:00)
Awarding Time: 2023-06-24 (time to be determined)