vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 17:14:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

remove quick lowvram option

Browse source
This commit is contained in:
Concedo 2024-05-02 14:21:44 +08:00
parent fb7e72352e
commit 0d8c4a9b73
1 changed files with 0 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

3
koboldcpp.py
View file

@ -1987,7 +1987,6 @@ def show_new_gui():
if index == "Use CuBLAS" or index == "Use hipBLAS (ROCm)":
lowvram_box.grid(row=4, column=0, padx=8, pady=1, stick="nw")
quick_lowvram_box.grid(row=4, column=0, padx=8, pady=1, stick="nw")
mmq_box.grid(row=4, column=1, padx=8, pady=1, stick="nw")
quick_mmq_box.grid(row=4, column=1, padx=8, pady=1, stick="nw")
splitmode_box.grid(row=5, column=1, padx=8, pady=1, stick="nw")
@ -1995,7 +1994,6 @@ def show_new_gui():
tensor_split_entry.grid(row=8, column=1, padx=8, pady=1, stick="nw")
else:
lowvram_box.grid_forget()
quick_lowvram_box.grid_forget()
mmq_box.grid_forget()
quick_mmq_box.grid_forget()
tensor_split_label.grid_forget()
@ -2033,7 +2031,6 @@ def show_new_gui():
quick_gpuname_label.grid(row=3, column=1, padx=75, sticky="W")
quick_gpuname_label.configure(text_color="#ffff00")
quick_gpu_layers_entry,quick_gpu_layers_label = makelabelentry(quick_tab,"GPU Layers:", gpulayers_var, 6, 50,"How many layers to offload onto the GPU.\nVRAM intensive, usage increases with model and context size.\nRequires some trial and error to find the best fit value.")
quick_lowvram_box = makecheckbox(quick_tab, "Low VRAM (No KV offload)", lowvram_var, 4,0,tooltiptxt="Avoid offloading KV Cache or scratch buffers to VRAM.\nAllows more layers to fit, but may result in a speed loss.")
quick_mmq_box = makecheckbox(quick_tab, "Use QuantMatMul (mmq)", mmq_var, 4,1,tooltiptxt="Enable MMQ mode instead of CuBLAS for prompt processing. Read the wiki. Speed may vary.")