Faraday LogoFaraday.dev

Desktop AppCharacter Hub
Learning CenterBlogChangelogCompany
We're hiring

Faraday Logo

Latest Updates & Improvements to Faraday
Feb 28, 2024
  • Desktop-to-Mobile Tethering
    • Connect to your Faraday desktop app from your mobile browser
    • All data is stored locally on your computer
    • Default Off - Can be toggled on and off in the settings
    • Learn more: docs.faraday.dev/tethering
  • Other fixes & improvements
    • Better VRAM handling on MacOS
    • Better error messages when the model fails to start on Windows
Feb 21, 2024
  • More improvements to the Windows GPU auto-detection system
    • Fixed issue where some GPUs would not be detected
    • Fixed issue where some GPUs would be detected but not selectable
    • Fixed various issues with Vulkan-based GPU acceleration
  • Allow for arbitrary maximum context length in Advanced Settings
  • Fix colons appearing randomly in the output
0.15.0 - 0.15.4 Internal
Feb 13, 2024
  • Small bug fixes and improvements
    • Fixed issue causing crashing when using Vulkan-based GPU acceleration
    • Fixed small bug causing messages to show twice while waiting for the model to start
    • Fixed images showing up as square when posting to Character Hub
    • Fixed issue where long lines in codeblocks would overflow the chat window
    • Fixed issue where blinking cursor was not showing up while generating text
0.14.1 - 0.14.3 Internal
Feb 12, 2024
  • Improvements to model startup logic
    • Models no longer unload when leaving the chat page
    • Model is reused when switching to a different Character with the same settings
    • Better loading messages during model startup
    • Various performance improvements and fixes to GPU-based inference
  • Added syntax highlighting to Codeblocks on the chat page
    • use ``` (three backticks) to start and end a codeblock
    • Codeblocks can now streamed in real-time
    • Also fixed issue where chat message would flicker with "quotes", *actions*, inline code, and codeblocks
  • Various other small bugs and UI issues
    • Fix: issue where mobile chats would not scroll to bottom of the page on load
    • Fix: flickering during text generation with asterisks and quotes
  • "Experimental backend" update - please help us test and give us feedback on Discord!
    • Support for SOTA 2-bit quants
    • Support for Q3_K_XS
    • Cuda/Vulkan/Metal performance improvements
Feb 6, 2024
  • Added support for Vulkan-based GPU acceleration
    • This is a beta feature and may not work on all devices - support can be toggled in the settings
    • Faster token generation rates on AMD GPUs
  • Fix: Various small UI issues and typos
  • New experimental backend - can be toggled in the settings
0.13.18 - 0.13.19 Internal
Jan 31, 2024
  • Complete rewrite of the Windows GPU auto-detection system
    • Faster model startup times on all Windows devices
    • Faster token generation rates on some Windows devices
    • Better detection of VRam leading to significantly better utilization of the GPU
  • Support for ChatML-based character prompts
    • Only recommended for advanced users
    • Model metadata will be updated to support this feature in the coming days
  • Other fixes & improvements
    • Better error & loading messages across the app
    • Updated trending algorithm on the Character Hub
    • Fix: Issue where the temperature could not be set above 2.0
0.13.15 - 0.13.16 Internal
Jan 17, 2024
  • Text To Speech is live and free for everyone!
    • Go to a character's settings to turn it on
    • Three male and three female voices to choose from
    • Control talking speed, auto-play, and input filters
  • Better Search & Sorting on the Character Hub
    • Faster search across the board (2x to 10x speedup)
    • Fuzzy search support
    • Added "Trending" & "Newest" pages to the sidebar
    • Added sorting to all hub pages
  • Other fixes & improvements
    • Separate vRAM and RAM memory usage in UI
    • Stability fixes for Cloud Models
    • Fix: Timeout issue when uploading characters to the hub
    • Fix: Issue when lorebook items have * in the title
    • Fix Author's note text input blocking standard keyboard shortcuts
0.13.13 Internal
Jan 5, 2024
  • Rollback of 0.13.10 backend performance improvements causing issues on some Windows devices
0.13.11 Internal
Jan 3, 2024
  • Improved backend performance on most Windows devices (both CPU and GPU)
    • This includes faster model loading, improved token rates, and optimized memory usage
    • If you have had issues on auto-GPU mode since v0.13.1, please try again on this release
  • Fix: "Newest Characters" page on Hub missing option to display more results
  • Support for infinite-length chats
  • Text-To-Speech beta for subscribers to our Early Supporters plan
    • To activate this feature, you must sign into your Character Hub account on the Desktop app
  • Added UtopiaXL 13B and Psyonic-Cetacean 20B to Faraday Cloud
0.13.7 - 0.13.9 Internal
Dec 23, 2023
  • Replace Mirostat with Min-P sampling
  • 6144 model context option in settings
  • Fix for auto-GPU mode overshooting available vRAM for 8192 context size
  • Fix for "Following" section on the Hub showing random Characters
© 2023 Ahoy Labs, Inc.
Community Guidelines
Terms of Use
Privacy Policy
Character Hub
Tethering: Disabled
No model loaded