vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 09:04:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

add additional tooltips (#710)

Browse source
This commit is contained in:
YellowRoseCx 2024-02-25 21:15:57 -06:00 • committed by GitHub
parent d47e13c892
commit 7b85917827
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
koboldcpp.py
View file

@ -1650,10 +1650,10 @@ def show_new_gui():
gpuname_label.grid(row=3, column=1, padx=75, sticky="W") gpuname_label.grid(row=3, column=1, padx=75, sticky="W")
gpuname_label.configure(text_color="#ffff00") gpuname_label.configure(text_color="#ffff00")
gpu_layers_entry,gpu_layers_label = makelabelentry(hardware_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.") gpu_layers_entry,gpu_layers_label = makelabelentry(hardware_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.")
tensor_split_entry,tensor_split_label = makelabelentry(hardware_tab, "Tensor Split:", tensor_split_str_vars, 8, 80) tensor_split_entry,tensor_split_label = makelabelentry(hardware_tab, "Tensor Split:", tensor_split_str_vars, 8, 80, tooltip='When using multiple GPUs this option controls how large tensors should be split across all GPUs.\nUses a comma-separated list of non-negative values that assigns the proportion of data that each GPU should get in order.\nFor example, "3,2" will assign 60% of the data to GPU 0 and 40% to GPU 1.')
lowvram_box = makecheckbox(hardware_tab, "Low VRAM", lowvram_var, 4,0) lowvram_box = makecheckbox(hardware_tab, "Low VRAM", lowvram_var, 4,0, tooltiptxt='Select lowvram to not allocate VRAM scratch buffer/ "K" & "V" Cache.\nCan save a decent amount of VRAM, but makes fully offloaded model processing a bit slower.')
mmq_box = makecheckbox(hardware_tab, "Use QuantMatMul (mmq)", mmq_var, 4,1) mmq_box = makecheckbox(hardware_tab, "Use QuantMatMul (mmq)", mmq_var, 4,1, tooltiptxt="Enable MMQ mode to use finetuned kernels instead of default CuBLAS/HipBLAS for prompt processing.\nRead the wiki. Speed may vary.")
splitmode_box = makecheckbox(hardware_tab, "Row-Split", rowsplit_var, 5,0) splitmode_box = makecheckbox(hardware_tab, "Row-Split", rowsplit_var, 5,0, tooltiptxt="Split rows across GPUs instead of splitting layers and KV across GPUs.\nUses the main GPU for small tensors and intermediate results. Speed may vary.")
# threads # threads
makelabelentry(hardware_tab, "Threads:" , threads_var, 11, 50,"How many threads to use.\nRecommended value is your CPU core count, defaults are usually OK.") makelabelentry(hardware_tab, "Threads:" , threads_var, 11, 50,"How many threads to use.\nRecommended value is your CPU core count, defaults are usually OK.")