Summer is approaching and autumn is approaching, forest clocks and summer shifts, ModelWhale is ushering in a new round of version updates, looking forward to bringing you a better user experience.
In this update, ModelWhale mainly performs the following functional iterations:
• Added portal content to support self-disclosure by members (team version ✓)
• New iterative version switching of portal content (team version ✓)
• Optimize projects, data search and filtering (Professional version ✓ Team version ✓ )
• Added the ability to switch computing power specifications during runtime (basic version ✓ professional version ✓ team version ✓ )
• Added user API token access credentials (team version ✓)
1. Added portal content to support self-disclosure by members (team version ✓)
Open data and sharing of results can play a greater value. While ModelWhale provides data science research tools, it also provides each team with an "independent portal" as a content sharing window to connect with external partners. In addition to the administrator being able to publish selected content within the organization to the portal, organization members are now also supported to apply for disclosure independently to realize the rapid construction of portal content. Taking organization data as an example, members click "Publish to Portal", complete the specification overview and submit the application; after the organization administrator approves, the content disclosure is completed.
Tips: We also provide a "whitelist" function. Whitelist users can directly display their content (data, code items) to the portal without review.
2. New iterative version switching of portal content (team version ✓)
The ModelWhale code project supports "version management" to help users record and trace each update iteration. When the content of the code project is updated and iterated, and the portal content needs to be adjusted accordingly, the user can "switch the version" by himself (no need to apply for content disclosure again), and the organization administrator can also operate the "switch version" in the management background.
3. Optimizing projects, data search and filtering (professional version ✓ team version ✓ )
When the problems we want to solve become complex, it is inevitable that a lot of content will be generated in the process. ModelWhale projects and data now support more powerful search and filter functions: (1) The search scope includes all "projects and data created by myself" and "projects and data shared by others"; (2) support by name, Introduction, overview, metadata fields, search for related content; (3) You can also filter the required content by tags, creators, license agreements, and time.
4. Added the ability to switch computing power specifications during runtime (basic version ✓ professional version ✓ team version ✓)
ModelWhale supports flexible scheduling of computing power of different specifications and convenient scaling and switching. You can use CPU computing power for code debugging and method inspection, and then switch to high-spec CPU and GPU computing power for full data processing and model training. In the upper right corner of the ModelWhale Notebook, we provide an entry to quickly switch resources to help you achieve this goal faster: when switching computing power, you can also switch the image at the same time, and let the platform build a new cloud operating environment for you, satisfying the following A step-by-step analysis of needs.
Tips: For long-running computing tasks and model training, it is recommended to use the "offline task" function for running hosting (at the same time, use GPU cluster computing power for distributed computing); see the offline task manual for more details .
5. Add user API token access credentials (team version ✓)
Users can generate API token access credentials as needed, and call the interface in ModelWhale through the token: After clicking "Generate Access Credentials", we will generate a string of token characters for you. After closing the pop-up window, you cannot access it again. It is recommended to save it properly. You can set the name of the access credential, the expiration time, and revoke the existing access credential.
The above is the whole content of the ModelWhale version update in this issue.
Enter ModelWhale.com , try the professional version (individual research) or the team version (organizational collaboration) for free, and get free CPU and GPU computing power! (It is recommended to use the computer for trial experience)
If you have any suggestions, questions about ModelWhale, or need to renew the trial, please contact MW , MoMo is happy to serve and communicate with you.