refactored a lot of code, remove bantokens, move it to api

This commit is contained in:
Concedo 2024-04-27 17:57:13 +08:00
parent 4ec8a9c57b
commit c230b78906
6 changed files with 214 additions and 76 deletions

View file

@ -270,7 +270,7 @@ class model_backend(InferenceModel):
port=5001, port_param=5001, host='', launch=False, lora=None, threads=self.kcpp_threads, blasthreads=self.kcpp_threads,
psutil_set_threads=False, highpriority=False, contextsize=self.kcpp_ctxsize,
blasbatchsize=self.kcpp_blasbatchsize, ropeconfig=[self.kcpp_ropescale, self.kcpp_ropebase], stream=False, smartcontext=self.kcpp_smartcontext,
unbantokens=False, bantokens=None, usemirostat=None, forceversion=0, nommap=self.kcpp_nommap,
usemirostat=None, forceversion=0, nommap=self.kcpp_nommap,
usemlock=False, noavx2=self.kcpp_noavx2, debugmode=self.kcpp_debugmode, skiplauncher=True, hordeconfig=None, noblas=self.kcpp_noblas,
useclblast=self.kcpp_useclblast, usecublas=self.kcpp_usecublas, usevulkan=self.kcpp_usevulkan, gpulayers=self.kcpp_gpulayers, tensor_split=self.kcpp_tensor_split, config=None,
onready='', multiuser=False, foreground=False, preloadstory=None, noshift=False, remotetunnel=False, ssl=False, benchmark=None, nocertify=False, sdconfig=None, mmproj=None,