Rollout(v5.0.0): Major version rollout

We've revamped our search functionality to provide more accurate and relevant results. Users can now search for specific keywords, tags, and categories, making it easier to find what they're looking for. Added many new accessibility features, including keyboard navigation and screen reader support, to ensure that everyone can use our platform with ease. Few visual bugs like on brave and chrome have been fixed. This version has significant improvements and is ready for everyone to try out. We'll be rolling out specific beta urls to beta testers as well as stack API for public beta.

New Feature(v4.9.1): Settings Options

Users can now customize their experience with our new settings options. From changing themes to adjusting profile icon preferences, we've made it easier for users to tailor their experience to their needs.

Bug Fix(v4.8.0): Resolved Issues

We've addressed several issues with the comment section, Giscus, and AI fetches including problems with comment threading and reply functionality. We've fixed the fetching function and now have multiple servers which are choosen dynamically, application will receive a response from the server with the least load. And more things are now handled on server Users should now experience a better overall experience with these features.

Performance Optimization(v4.5.0)

We've implemented several performance optimizations to improve the overall speed and responsiveness of the platform. Users should notice faster page loads and improved navigation. We've also added a redirection mechanism to ensure seamless functionality. This ensures personalisation of UI elements based on user preferences and behaviors.

Release v4.0

Release v4.0 focuses on optimization and improving user expreience. Our newest flaskapi-chi which will enable less load on client side. Testing for a mechanism where data is fetched rather than storing on server has begun. Anti-plagarism methods have been introduced where no one can use our framework and publish it as theres. So we can keep code opensource for people who want to know how our systems work rather thancopy pasting our code for there own profits.

Rebranded(v3.8.4 > v3.9)

We are excited to announce that Proximity.gg has been rebranded as Vicinity.fr. This change reflects our commitment to enhancing our brand identity and expanding our services to better meet the needs of our global community. Below are the detailed updates and improvements associated with this rebranding.

Addition(v3.8.3): Read accessibility

Read More / Read Less Feature: Implemented a new "Read More / Read Less" feature to improve content accessibility and user experience. Users can now expand and collapse lengthy content sections for easier reading. Close About Dialogue: Added functionality to close the About dialogue box, enhancing the overall navigational flow and user control over the interface.

Addition (v3.8.2): Firescore Method Tooltip

The fire scores are now calculated using a new AI model by mgalkin, the mgalkin/ultra_50g model. There are three types of estimations for Firescores:

• Estimation: When certain parameters are passed to the AI model to get the Firescores.
• Real-time: Firescores are calculated in real-time as all parameters are passed.
• Quasi Estimation: Additional params and information are processed through the AI to get a highly precise
• Note: These are not always consistent as the same Firescore are now calculated differently due to varying methods.

Addition(v3.8.2): Server Status Notifier

Added new server status indicators to provide detailed real-time updates on server conditions. Each status is associated with a specific color code and descriptive message.

Addition(v3.8.1): A Next Battle Notifier

Battle switch places if it is guaranteed that, 2nd place will take on 1st just after being dethroned but if it takes a long time (f>24hours) or infinite time (i.e. never) then new battle starts between new users.

Addition(v3.8.0): Notifier Icons

The fire scores are calculated by new AI model by mgalkin, the mgalkin/ultra_50g model. But during the preview phase, it was observed that it sometimes would not calculate all fire scores or incomplete parameters would be sent to the model resulting in inaccurate scores, so it uses the old model to estimate the score. '🔄' : 'Estimated Score' & '🔹' : 'Realtime Score'

Addition(v3.7.9): Scorefire Color Indicators

Introduced new scorefires(F) and scorebadges(B) to better represent the types of growth observed in user scores:
Greenf (●): Indicates stable growth.
Bluef (●): Indicates very fast growth.
Orangef (●): Indicates growth.
Pinkb (●): Indicates slow growth.
Redb (●): Indicates declining graph.
Grayb (●): Indicates stagnant graph.

Removal(v3.7.9): Depriciated Old Model

The Graphormer model clefourrier/graphormer-base-pcqm4mv1 is ressource intensive for large graphs, and might lead to OOM errors. So thid old model has been completely depriciated and the new AI model by mgalkin, the mgalkin/ultra_50g model is now the default model.

Bug Fix(v3.7.7): Alignment fixed

Fixed minor bugs related to the display of profile images and the alignment of text within the leaderboard entries.

Optimization(v3.7.6): Performance Tuning

Optimized memory usage and reduced app footprint. Implemented caching mechanisms for faster data retrieval. Conducted load testing to ensure scalability under high user traffic.

Update (v3.7.6): User Tooltip

The ShynBui/comment_classification_v2 model is used to generate tooltips based on parameters that are updated every 100 hours. This model is currently incompatible with our system, but the update frequency will be reduced further. Tooltips will appear when hovered over.

Addition(v3.7.6): Method Changed

Resolved issues related to occasional crashes or errors. Fixed bugs affecting prediction accuracy under specific scenarios. Conducted thorough regression testing to validate fixes and parallelize the process by using a new model on the side.

Bug Fix(v3.7.5): Alignment fixed

Fixed minor bugs related to the display of profile images and the alignment of text within the leaderboard entries.

Bug Fix(v3.7.5): Data Loss

When the app updates, live scores becomes corrupted and passes flawed parameters, resulting in inaccurate growth calculations leading to sometimes crashes or just errors. Fail-safe measures are implimented which aim to break the loop and prevent crashes. We will solve this issue completely or change the model used.

Addition(v3.7.4): User Dropdown

Just a new accessibility feature will add the proximity pro feature sometime later.