vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

more granular context size selections

Browse source
This commit is contained in:
Concedo 2023-07-01 11:02:44 +08:00
parent eda663f15f
commit 632bf27b65
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
koboldcpp.py
View file

@ -859,7 +859,7 @@ if __name__ == '__main__':
parser.add_argument("--blasthreads", help="Use a different number of threads during BLAS if specified. Otherwise, has the same value as --threads",metavar=('[threads]'), type=int, default=0) parser.add_argument("--blasthreads", help="Use a different number of threads during BLAS if specified. Otherwise, has the same value as --threads",metavar=('[threads]'), type=int, default=0)
parser.add_argument("--psutil_set_threads", help="Experimental flag. If set, uses psutils to determine thread count based on physical cores.", action='store_true') parser.add_argument("--psutil_set_threads", help="Experimental flag. If set, uses psutils to determine thread count based on physical cores.", action='store_true')
parser.add_argument("--highpriority", help="Experimental flag. If set, increases the process CPU priority, potentially speeding up generation. Use caution.", action='store_true') parser.add_argument("--highpriority", help="Experimental flag. If set, increases the process CPU priority, potentially speeding up generation. Use caution.", action='store_true')
parser.add_argument("--contextsize", help="Controls the memory allocated for maximum context size, only change if you need more RAM for big contexts. (default 2048)", type=int,choices=[512,1024,2048,4096,8192], default=2048) parser.add_argument("--contextsize", help="Controls the memory allocated for maximum context size, only change if you need more RAM for big contexts. (default 2048)", type=int,choices=[512,1024,2048,3072,4096,6144,8192], default=2048)
parser.add_argument("--blasbatchsize", help="Sets the batch size used in BLAS processing (default 512). Setting it to -1 disables BLAS mode, but keeps other benefits like GPU offload.", type=int,choices=[-1,32,64,128,256,512,1024], default=512) parser.add_argument("--blasbatchsize", help="Sets the batch size used in BLAS processing (default 512). Setting it to -1 disables BLAS mode, but keeps other benefits like GPU offload.", type=int,choices=[-1,32,64,128,256,512,1024], default=512)
parser.add_argument("--stream", help="Uses streaming when generating tokens. Only for the Kobold Lite UI.", action='store_true') parser.add_argument("--stream", help="Uses streaming when generating tokens. Only for the Kobold Lite UI.", action='store_true')
parser.add_argument("--smartcontext", help="Reserving a portion of context to try processing less frequently.", action='store_true') parser.add_argument("--smartcontext", help="Reserving a portion of context to try processing less frequently.", action='store_true')