diff --git a/koboldcpp.py b/koboldcpp.py index 5ee1a5031..56a1589b9 100644 --- a/koboldcpp.py +++ b/koboldcpp.py @@ -2512,17 +2512,17 @@ def show_gui(): # quick boxes quick_boxes = { - "Launch Browser": {"variable": launchbrowser, "description": "Launches your default browser after model loading is complete"}, - "Disable MMAP": {"variable": disablemmap, "description": "Avoids using mmap to load models if enabled"}, - "Use ContextShift": {"variable": contextshift, "description": "Uses Context Shifting to reduce reprocessing.\nRecommended. Check the wiki for more info."}, - "Remote Tunnel": {"variable": remotetunnel, "description": "Creates a trycloudflare tunnel.\nAllows you to access koboldcpp from other devices over an internet URL."}, - "Use FlashAttention": {"variable": flashattention, "description": "Enable flash attention for GGUF models."}, - "Quiet Mode": {"variable": quietmode, "description": "Prevents all generation related terminal output from being displayed."} + "Launch Browser": [launchbrowser, "Launches your default browser after model loading is complete"], + "Disable MMAP": [disablemmap, "Avoids using mmap to load models if enabled"], + "Use ContextShift": [contextshift, "Uses Context Shifting to reduce reprocessing.\nRecommended. Check the wiki for more info."], + "Remote Tunnel": [remotetunnel, "Creates a trycloudflare tunnel.\nAllows you to access koboldcpp from other devices over an internet URL."], + "Use FlashAttention": [flashattention, "Enable flash attention for GGUF models."], + "Quiet Mode": [quietmode, "Prevents all generation related terminal output from being displayed."] } - + for idx, (name, properties) in enumerate(quick_boxes.items()): - makecheckbox(quick_tab, name, properties["variable"], int(idx/2) + 20, idx % 2, tooltiptxt=properties["description"]) - + makecheckbox(quick_tab, name, properties[0], int(idx/2) + 20, idx % 2, tooltiptxt=properties[1]) + # context size makeslider(quick_tab, "Context Size:", contextsize_text, context_var, 0, len(contextsize_text)-1, 30, width=280, set=5,tooltip="What is the maximum context size to support. Model specific. You cannot exceed it.\nLarger contexts require more memory, and not all models support it.") @@ -2559,17 +2559,17 @@ def show_gui(): # hardware checkboxes hardware_boxes = { - "Launch Browser": {"variable": launchbrowser, "description": "Launches your default browser after model loading is complete"}, - "High Priority": {"variable": highpriority, "description": "Increases the koboldcpp process priority.\nMay cause lag or slowdown instead. Not recommended."}, - "Disable MMAP": {"variable": disablemmap, "description": "Avoids using mmap to load models if enabled"}, - "Use mlock": {"variable": usemlock, "description": "Enables mlock, preventing the RAM used to load the model from being paged out."}, - "Debug Mode": {"variable": debugmode, "description": "Enables debug mode, with extra info printed to the terminal."}, - "Keep Foreground": {"variable": keepforeground, "description": "Bring KoboldCpp to the foreground every time there is a new generation."} + "Launch Browser": [launchbrowser, "Launches your default browser after model loading is complete"], + "High Priority": [highpriority, "Increases the koboldcpp process priority.\nMay cause lag or slowdown instead. Not recommended."], + "Disable MMAP": [disablemmap, "Avoids using mmap to load models if enabled"], + "Use mlock": [usemlock, "Enables mlock, preventing the RAM used to load the model from being paged out."], + "Debug Mode": [debugmode, "Enables debug mode, with extra info printed to the terminal."], + "Keep Foreground": [keepforeground, "Bring KoboldCpp to the foreground every time there is a new generation."] } - + for idx, (name, properties) in enumerate(hardware_boxes.items()): - makecheckbox(hardware_tab, name, properties["variable"], int(idx/2) + 30, idx % 2, tooltiptxt=properties["description"]) - + makecheckbox(hardware_tab, name, properties[0], int(idx/2) + 30, idx % 2, tooltiptxt=properties[1]) + # blas thread specifier makelabelentry(hardware_tab, "BLAS threads:" , blas_threads_var, 14, 50,tooltip="How many threads to use during BLAS processing.\nIf left blank, uses same value as regular thread count.") # blas batch size