vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 09:34:37 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

update license, added backwards compatibility with both ggml model formats, fixed context length issues.

Browse source
This commit is contained in:
Concedo 2023-03-20 23:43:35 +08:00
parent a2c10e0d2f
commit 8d39365af6
11 changed files with 807 additions and 15 deletions
Download patch file Download diff file Expand all files Collapse all files

5
llama_for_kobold.py
View file

@ -74,8 +74,8 @@ import json, http.server, threading, socket, sys, time
# global vars
global friendlymodelname
friendlymodelname = ""
maxctx = 1024
maxlen = 256
maxctx = 512
maxlen = 128
modelbusy = False
port = 5001
last_context = ""
@ -151,6 +151,7 @@ class ServerRequestHandler(http.server.BaseHTTPRequestHandler):
if last_context!="" and newprompt.startswith(last_context):
fresh_state = False
newprompt = newprompt[len(last_context):]
print("Resuming state, new input len: " + str(len(newprompt)))
#print("trimmed: " + newprompt)
recvtxt = generate(
prompt=newprompt,