v1.2.0

improve details
officially support macOS and Linux

Install

Windows: https://github.com/josStorer/RWKV-Runner/blob/master/build/windows/Readme_Install.txt
MacOS: https://github.com/josStorer/RWKV-Runner/blob/master/build/darwin/Readme_Install.txt
Linux: https://github.com/josStorer/RWKV-Runner/blob/master/build/linux/Readme_Install.txt

v1.1.9

OpenFileFolder for Linux
add ResetConfigsButton to Home Page
Improved cross-platform interaction
Improve development config

Improve the built-in download function, enhance the logic robustness and reliability in adverse network environments to avoid the problem of PytorchStreamReader failed reading zip archive: invalid header or archive is corrupted, as well as the problem of loading corrupted files leading to garbled answers.
Currently, it has been confirmed through the following tests that the SHA256 of the downloaded files are consistent after completion of the download: select a model for download, continuously switch network proxies during the download process, constantly click pause and resume, unplug the network cable, force quit the program, and the integrity of the downloaded file is still normal upon completion.

v1.1.8

fix the crash issue caused by temperature being 0
update logo

v1.1.7

improve logs
display current strategy
add strategy guides
remove langchain from dep_check

v1.1.6

support MX, Tesla P, Quadro P, NVIDIA P, TITAN X, RTX A series, TITAN RTX and RTX TITAN Ada
add api logs

v1.1.5

support macOS (currently you have to compile it by yourself)
move state cache to memory to avoid vram error
add Instruction to Completion Presets
improve user guides (vc++ installation, path contains space, administrator privileges)

v1.1.4

fix_tokens
fix Cmd and CopyFile

v1.1.3

custom models path (#28)
custom python path (#26 #28)
custom strategy mode
improve system for rwkv-4-world
add button to reset all configs
tesla P104 P106 kernel
safe update-config
other improvements

v1.1.2

fix downloads page

v1.1.1

embed all core dependencies
support for tesla P40 custom cuda kernel
many details improvements
allow system (chat API) to be placed anywhere
update models source and hide old models