Skip to content

Releases: josStorer/RWKV-Runner

v1.5.0

05 Nov 13:27
Compare
Choose a tag to compare

Changes

Features

  • chat attachment is now related to single message (Experimental)
  • port occupied detection

Upgrades

  • upgrade to rwkv 0.8.20

Improvements

  • improve the compatibility between frontend presets and chatgpt api
  • improve memory usage of state cache

Chores

  • update ngrok_connect
  • python38 compatibility
  • adjust startup process

Fixes

  • fix log encoding error
  • fix stop button status of Chat page

Install

v1.4.9

27 Oct 06:28
Compare
Choose a tag to compare

Changes

Features

  • allow conversation with some document (.pdf, .txt) (Experimental)
  • add /file-to-text api
  • allow avatarImg to be local absolute path
  • base64 preset support

Upgrades

  • upgrade to rwkv 0.8.16 (DirectML support; rwkv 5.2 no longer needs to ensure custom cuda kernel enabled)
  • upgrade to webgpu 0.2.2 (WebGPU Mode is now recommended for AMD and Intel Users) (https://github.com/josStorer/ai00_rwkv_server)
  • upgrade python packages

Improvements

  • improve cuda kernel compatibility (compute compatibility 5.3, Jetson Nano, MX Series, Nvidia 10 Series+)
  • RWKVType now no longer relies on the file name (use emb)
  • improve message interruption and retry for Chat page
  • update sample.jsonl of LoRA finetune
  • update api stop strategy for better custom user_name and assistant_name support
  • edited chat message now is marked as Normal
  • change default World series prefix to User/Assistant

Chores

  • update manifest.json (RWKV-5)
  • update readme and client text description
  • add pip --no-warn-script-location
  • mark rwkv raven series as old model
  • chore

Fixes

  • fix linux kernel (partial revert 68228a4)
  • fix the make command on Linux and macOS, no longer need manual operations on the wsl.go file. (#158, #173, #207)

Install

v1.4.8

03 Oct 07:21
Compare
Choose a tag to compare

v1.4.7

18 Sep 15:20
Compare
Choose a tag to compare

Changes

  • custom tokenizer .txt support
  • add pre-compiled beta cuda kernel (rwkv-beta==0.8.5, 40%+ faster for fp16) (thanks to #180, pre-compiled kernel of RTX 40 Series will be included later)

Install

v1.4.6

16 Sep 05:23
Compare
Choose a tag to compare

Changes

  • frontend adaptation for api params (user_name, assistant_name, presystem)
  • custom tokenizer (#77)
  • enable right-click context menu
  • upgrade cuda-beta
  • revert(2d5456): refresh local models when download complete (for macOS)
  • improve ui desc
  • chore

Install

v1.4.5

27 Aug 16:15
Compare
Choose a tag to compare

Changes

  • frontend: update manifest (a lot of new models)
  • frontend: correct Preset UI description
  • frontend: add HardwareMonitor (Windows Only)
  • lora finetune: fix max_epochs (#170)
  • python-backend: allow message content to be empty
  • python-backend: extra ChatCompletionBody params (raw, presystem)
  • python-backend: add default_stop when stop is null
  • webgpu: fix webgpu_server file permissions of linux and macos
  • chore

Install

v1.4.4

16 Aug 15:46
Compare
Choose a tag to compare

Changes

  • webgpu support (AMD, Intel, Nvidia, Apple) (https://github.com/cgisky1980/ai00_rwkv_server)
  • add rwkv-cuda-beta support (faster)
  • add misc API (/models and /dashboard/billing/credit_grants)
  • allow multiple systems
  • allow completions input to be null

Install

WebGPU Mode

image

v1.4.3

31 Jul 15:03
Compare
Choose a tag to compare

v1.4.2

29 Jul 11:43
Compare
Choose a tag to compare

Breaking Changes

Due to performance and bandwidth considerations, the /chat/completions and /completions API no longer return the response field. If necessary, you can still uncomment it yourself.
Please note that this is never part of the OpenAI API, it existed previously only for API development convenience. If you follow the OpenAI API specification, you will not be affected in any way.

Changes

  • improve /chat/completions and /completions API performance (remove response field)
  • improve default ChatCompletion stop
  • improve python backend startup speed
  • update defaultConfigs

Install

v1.4.1

28 Jul 14:28
Compare
Choose a tag to compare

Changes

  • add Composition Page (RWKV-Music)
  • improve RunButton prompt
  • support for stop array api params
  • improve embeddings API results
  • improve python backend startup speed
  • add support for MIDI RWKV
  • add midi api
  • add CPU-120M-Music config
  • improve sse fetch
  • update manifest (a lot of new models)
  • update presets
  • remove LoraFinetunePrecision fp32
  • chore

Install