vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-05-22 03:10:03 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

more consistency fixes

Browse source
This commit is contained in:
Concedo 2025-08-13 19:28:53 +08:00
parent 955cf66bbc
commit 4b2ca1169c
3 changed files with 5 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
koboldcpp.py
View file

@ -1898,7 +1898,7 @@ def embeddings_load_model(model_filename):
inputs = embeddings_load_model_inputs()
inputs.model_filename = model_filename.encode("UTF-8")
inputs.gpulayers = (999 if args.embeddingsgpu else 0)
inputs.flash_attention = False
inputs.flash_attention = args.flashattention
inputs.threads = args.threads
inputs.use_mmap = args.usemmap
inputs.embeddingsmaxctx = (args.embeddingsmaxctx if args.embeddingsmaxctx else args.contextsize) # for us to clamp to contextsize if embeddingsmaxctx unspecified