vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2026-07-10 01:18:32 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

added v1 voices endpoint, added lcpp aliases for cli, fixed dia wrong voice

Browse source
This commit is contained in:
Concedo 2025-08-30 11:20:18 +08:00
parent 645b09ea20
commit 7b396bd917
3 changed files with 43 additions and 21 deletions
Download patch file Download diff file Expand all files Collapse all files

20
kcpp_docs.embd
View file

@ -438,6 +438,26 @@
"quiet": {
"type": "boolean",
"description": "Is the server in quiet (no logging) mode."
},
"last_input_count": {
"type": "integer",
"description": "Number of input tokens in the last request."
},
"last_process_time": {
"type": "number",
"description": "Process time for last request."
},
"last_eval_time": {
"type": "number",
"description": "Generation time for last request."
},
"last_process_speed": {
"type": "number",
"description": "Processing speed (tokens/sec)."
},
"last_eval_speed": {
"type": "number",
"description": "Generation speed (tokens/sec)."
}
},
"required": [

42
koboldcpp.py
View file

@ -3359,6 +3359,8 @@ Change Mode<br>
response_body = (json.dumps(["kobo","cheery","sleepy","shouty","chatty"]).encode()) #some random voices for them to enjoy
elif self.path.endswith('/speakers'): #xtts compatible
response_body = (json.dumps([{"name":"kobo","voice_id":"kobo","preview_url":""},{"name":"cheery","voice_id":"cheery","preview_url":""},{"name":"sleepy","voice_id":"sleepy","preview_url":""},{"name":"shouty","voice_id":"shouty","preview_url":""},{"name":"chatty","voice_id":"chatty","preview_url":""}]).encode()) #some random voices for them to enjoy
elif self.path.endswith('/v1/audio/voices'):
response_body = (json.dumps({"status":"ok","voices":["kobo","cheery","sleepy","shouty","chatty"]}).encode()) #some random voices for them to enjoy
elif self.path.endswith('/get_tts_settings'): #xtts compatible
response_body = (json.dumps({"temperature":0.75,"speed":1,"length_penalty":1,"repetition_penalty":1,"top_p":1,"top_k":4,"enable_text_splitting":True,"stream_chunk_size":100}).encode()) #some random voices for them to enjoy
@ -7561,7 +7563,7 @@ if __name__ == '__main__':
parser = argparse.ArgumentParser(description=f'KoboldCpp Server - Version {KcppVersion}')
modelgroup = parser.add_mutually_exclusive_group() #we want to be backwards compatible with the unnamed positional args
modelgroup.add_argument("--model", metavar=('[filenames]'), help="Model file to load. Accepts multiple values if they are URLs.", type=str, nargs='+', default=[])
modelgroup.add_argument("--model","-m", metavar=('[filenames]'), help="Model file to load. Accepts multiple values if they are URLs.", type=str, nargs='+', default=[])
modelgroup.add_argument("model_param", help="Model file to load (positional)", nargs="?")
portgroup = parser.add_mutually_exclusive_group() #we want to be backwards compatible with the unnamed positional args
portgroup.add_argument("--port", metavar=('[portnumber]'), help=f"Port to listen on. (Defaults to {defaultport})", default=defaultport, type=int, action='store')
@ -7569,39 +7571,39 @@ if __name__ == '__main__':
parser.add_argument("--host", metavar=('[ipaddr]'), help="Host IP to listen on. If this flag is not set, all routable interfaces are accepted.", default="")
parser.add_argument("--launch", help="Launches a web browser when load is completed.", action='store_true')
parser.add_argument("--config", metavar=('[filename]'), help="Load settings from a .kcpps file. Other arguments will be ignored", type=str, nargs=1)
parser.add_argument("--threads", metavar=('[threads]'), help="Use a custom number of threads if specified. Otherwise, uses an amount based on CPU cores", type=int, default=get_default_threads())
parser.add_argument("--threads","-t", metavar=('[threads]'), help="Use a custom number of threads if specified. Otherwise, uses an amount based on CPU cores", type=int, default=get_default_threads())
compatgroup = parser.add_mutually_exclusive_group()
compatgroup.add_argument("--usecuda", "--usecublas", "--usehipblas", help="Use CUDA for GPU Acceleration. Requires CUDA. Enter a number afterwards to select and use 1 GPU. Leaving no number will use all GPUs.", nargs='*',metavar=('[lowvram|normal] [main GPU ID] [mmq|nommq] [rowsplit]'), choices=['normal', 'lowvram', '0', '1', '2', '3', 'all', 'mmq', 'nommq', 'rowsplit'])
compatgroup.add_argument("--usevulkan", help="Use Vulkan for GPU Acceleration. Can optionally specify one or more GPU Device ID (e.g. --usevulkan 0), leave blank to autodetect.", metavar=('[Device IDs]'), nargs='*', type=int, default=None)
compatgroup.add_argument("--useclblast", help="Use CLBlast for GPU Acceleration. Must specify exactly 2 arguments, platform ID and device ID (e.g. --useclblast 1 0).", type=int, choices=range(0,9), nargs=2)
compatgroup.add_argument("--usecpu", help="Do not use any GPU acceleration (CPU Only)", 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 8192).",metavar=('[256 to 262144]'), type=check_range(int,256,262144), default=8192)
parser.add_argument("--gpulayers", help="Set number of layers to offload to GPU when using GPU. Requires GPU. Set to -1 to try autodetect, set to 0 to disable GPU offload.",metavar=('[GPU layers]'), nargs='?', const=1, type=int, default=-1)
parser.add_argument("--tensor_split", "--tensorsplit", help="For CUDA and Vulkan only, ratio to split tensors across multiple GPUs, space-separated list of proportions, e.g. 7 3", metavar=('[Ratios]'), type=float, nargs='+')
parser.add_argument("--contextsize","--ctx-size", "-c", help="Controls the memory allocated for maximum context size, only change if you need more RAM for big contexts. (default 8192).",metavar=('[256 to 262144]'), type=check_range(int,256,262144), default=8192)
parser.add_argument("--gpulayers","--gpu-layers","--n-gpu-layers","-ngl", help="Set number of layers to offload to GPU when using GPU. Requires GPU. Set to -1 to try autodetect, set to 0 to disable GPU offload.",metavar=('[GPU layers]'), nargs='?', const=1, type=int, default=-1)
parser.add_argument("--tensor_split","--tensorsplit","--tensor-split","-ts", help="For CUDA and Vulkan only, ratio to split tensors across multiple GPUs, space-separated list of proportions, e.g. 7 3", metavar=('[Ratios]'), type=float, nargs='+')
#more advanced params
advparser = parser.add_argument_group('Advanced Commands')
advparser.add_argument("--version", help="Prints version and exits.", action='store_true')
advparser.add_argument("--analyze", metavar=('[filename]'), help="Reads the metadata, weight types and tensor names in any GGUF file.", default="")
advparser.add_argument("--maingpu", help="Only used in a multi-gpu setup. Sets the index of the main GPU that will be used.",metavar=('[Device ID]'), type=int, default=-1)
advparser.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=[-1,16,32,64,128,256,512,1024,2048,4096], default=512)
advparser.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)
advparser.add_argument("--maingpu","--main-gpu","-mg", help="Only used in a multi-gpu setup. Sets the index of the main GPU that will be used.",metavar=('[Device ID]'), type=int, default=-1)
advparser.add_argument("--blasbatchsize","--batch-size","-b", 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=[-1,16,32,64,128,256,512,1024,2048,4096], default=512)
advparser.add_argument("--blasthreads","--threads-batch", help="Use a different number of threads during BLAS if specified. Otherwise, has the same value as --threads",metavar=('[threads]'), type=int, default=0)
advparser.add_argument("--lora", help="GGUF models only, applies a lora file on top of model.", metavar=('[lora_filename]'), nargs='+')
advparser.add_argument("--loramult", metavar=('[amount]'), help="Multiplier for the Text LORA model to be applied.", type=float, default=1.0)
advparser.add_argument("--noshift", help="If set, do not attempt to Trim and Shift the GGUF context.", action='store_true')
advparser.add_argument("--noshift","--no-context-shift", help="If set, do not attempt to Trim and Shift the GGUF context.", action='store_true')
advparser.add_argument("--nofastforward", help="If set, do not attempt to fast forward GGUF context (always reprocess). Will also enable noshift", action='store_true')
advparser.add_argument("--useswa", help="If set, allows Sliding Window Attention (SWA) KV Cache, which saves memory but cannot be used with context shifting.", action='store_true')
advparser.add_argument("--ropeconfig", help="If set, uses customized RoPE scaling from configured frequency scale and frequency base (e.g. --ropeconfig 0.25 10000). Otherwise, uses NTK-Aware scaling set automatically based on context size. For linear rope, simply set the freq-scale and ignore the freq-base",metavar=('[rope-freq-scale]', '[rope-freq-base]'), default=[0.0, 10000.0], type=float, nargs='+')
advparser.add_argument("--overridenativecontext", help="Overrides the native trained context of the loaded model with a custom value to be used for Rope scaling.",metavar=('[trained context]'), type=int, default=0)
compatgroup3 = advparser.add_mutually_exclusive_group()
compatgroup3.add_argument("--usemmap", help="If set, uses mmap to load model.", action='store_true')
advparser.add_argument("--usemlock", help="Enables mlock, preventing the RAM used to load the model from being paged out. Not usually recommended.", action='store_true')
advparser.add_argument("--usemlock","--mlock", help="Enables mlock, preventing the RAM used to load the model from being paged out. Not usually recommended.", action='store_true')
advparser.add_argument("--noavx2", help="Do not use AVX2 instructions, a slower compatibility mode for older devices.", action='store_true')
advparser.add_argument("--failsafe", help="Use failsafe mode, extremely slow CPU only compatibility mode that should work on all devices. Can be combined with useclblast if your device supports OpenCL.", action='store_true')
advparser.add_argument("--debugmode", help="Shows additional debug info in the terminal.", nargs='?', const=1, type=int, default=0)
advparser.add_argument("--onready", help="An optional shell command to execute after the model has been loaded.", metavar=('[shell command]'), type=str, default="",nargs=1)
advparser.add_argument("--benchmark", help="Do not start server, instead run benchmarks. If filename is provided, appends results to provided file.", metavar=('[filename]'), nargs='?', const="stdout", type=str, default=None)
advparser.add_argument("--prompt", metavar=('[prompt]'), help="Passing a prompt string triggers a direct inference, loading the model, outputs the response to stdout and exits. Can be used alone or with benchmark.", type=str, default="")
advparser.add_argument("--prompt","-p", metavar=('[prompt]'), help="Passing a prompt string triggers a direct inference, loading the model, outputs the response to stdout and exits. Can be used alone or with benchmark.", type=str, default="")
advparser.add_argument("--cli", help="Does not launch KoboldCpp HTTP server. Instead, enables KoboldCpp from the command line, accepting interactive console input and displaying responses to the terminal.", action='store_true')
advparser.add_argument("--genlimit","--promptlimit", help="Sets the maximum number of generated tokens, it will restrict all generations to this or lower. Also usable with --prompt or --benchmark.",metavar=('[token limit]'), type=int, default=0)
advparser.add_argument("--multiuser", help="Runs in multiuser mode, which queues incoming requests instead of blocking them.", metavar=('limit'), nargs='?', const=1, type=int, default=1)
@ -7616,16 +7618,16 @@ if __name__ == '__main__':
advparser.add_argument("--ssl", help="Allows all content to be served over SSL instead. A valid UNENCRYPTED SSL cert and key .pem files must be provided", metavar=('[cert_pem]', '[key_pem]'), nargs='+')
advparser.add_argument("--nocertify", help="Allows insecure SSL connections. Use this if you have cert errors and need to bypass certificate restrictions.", action='store_true')
advparser.add_argument("--mmproj", metavar=('[filename]'), help="Select a multimodal projector file for vision models like LLaVA.", default="")
advparser.add_argument("--mmprojcpu", help="Force CLIP for Vision mmproj always on CPU.", action='store_true')
advparser.add_argument("--mmprojcpu","--no-mmproj-offload", help="Force CLIP for Vision mmproj always on CPU.", action='store_true')
advparser.add_argument("--visionmaxres", metavar=('[max px]'), help="Clamp MMProj vision maximum allowed resolution. Allowed values are between 512 to 2048 px (default 1024).", type=int, default=default_visionmaxres)
advparser.add_argument("--draftmodel", metavar=('[filename]'), help="Load a small draft model for speculative decoding. It will be fully offloaded. Vocab must match the main model.", default="")
advparser.add_argument("--draftamount", metavar=('[tokens]'), help="How many tokens to draft per chunk before verifying results", type=int, default=default_draft_amount)
advparser.add_argument("--draftgpulayers", metavar=('[layers]'), help="How many layers to offload to GPU for the draft model (default=full offload)", type=int, default=999)
advparser.add_argument("--draftmodel","--model-draft","-md", metavar=('[filename]'), help="Load a small draft model for speculative decoding. It will be fully offloaded. Vocab must match the main model.", default="")
advparser.add_argument("--draftamount","--draft-max","--draft-n", metavar=('[tokens]'), help="How many tokens to draft per chunk before verifying results", type=int, default=default_draft_amount)
advparser.add_argument("--draftgpulayers","--gpu-layers-draft","--n-gpu-layers-draft","-ngld", metavar=('[layers]'), help="How many layers to offload to GPU for the draft model (default=full offload)", type=int, default=999)
advparser.add_argument("--draftgpusplit", help="GPU layer distribution ratio for draft model (default=same as main). Only works if multi-GPUs selected for MAIN model and tensor_split is set!", metavar=('[Ratios]'), type=float, nargs='+')
advparser.add_argument("--password", metavar=('[API key]'), help="Enter a password required to use this instance. This key will be required for all text endpoints. Image endpoints are not secured.", default=None)
advparser.add_argument("--ignoremissing", help="Ignores all missing non-essential files, just skipping them instead.", action='store_true')
advparser.add_argument("--chatcompletionsadapter", metavar=('[filename]'), help="Select an optional ChatCompletions Adapter JSON file to force custom instruct tags.", default="AutoGuess")
advparser.add_argument("--flashattention", help="Enables flash attention.", action='store_true')
advparser.add_argument("--flashattention","--flash-attn","-fa", help="Enables flash attention.", action='store_true')
advparser.add_argument("--quantkv", help="Sets the KV cache data type quantization, 0=f16, 1=q8, 2=q4. Requires Flash Attention for full effect, otherwise only K cache is quantized.",metavar=('[quantization level 0/1/2]'), type=int, choices=[0,1,2], default=0)
advparser.add_argument("--forceversion", help="If the model file format detection fails (e.g. rogue modified model) you can set this to override the detected format (enter desired version, e.g. 401 for GPTNeoX-Type2).",metavar=('[version]'), type=int, default=0)
advparser.add_argument("--smartcontext", help="Reserving a portion of context to try processing less frequently. Outdated. Not recommended.", action='store_true')
@ -7634,13 +7636,13 @@ if __name__ == '__main__':
advparser.add_argument("--exporttemplate", help="Exports the current selected arguments as a .kcppt template file", metavar=('[filename]'), type=str, default="")
advparser.add_argument("--nomodel", help="Allows you to launch the GUI alone, without selecting any model.", action='store_true')
advparser.add_argument("--moeexperts", metavar=('[num of experts]'), help="How many experts to use for MoE models (default=follow gguf)", type=int, default=-1)
advparser.add_argument("--moecpu", metavar=('[layers affected]'), help="Keep the Mixture of Experts (MoE) weights of the first N layers in the CPU. If no value is provided, applies to all layers.", nargs='?', const=999, type=int, default=0)
advparser.add_argument("--moecpu","--n-cpu-moe", "-ncmoe", metavar=('[layers affected]'), help="Keep the Mixture of Experts (MoE) weights of the first N layers in the CPU. If no value is provided, applies to all layers.", nargs='?', const=999, type=int, default=0)
advparser.add_argument("--defaultgenamt", help="How many tokens to generate by default, if not specified. Must be smaller than context size. Usually, your frontend GUI will override this.", type=check_range(int,64,8192), default=768)
advparser.add_argument("--nobostoken", help="Prevents BOS token from being added at the start of any prompt. Usually NOT recommended for most models.", action='store_true')
advparser.add_argument("--enableguidance", help="Enables the use of Classifier-Free-Guidance, which allows the use of negative prompts. Has performance and memory impact.", action='store_true')
advparser.add_argument("--maxrequestsize", metavar=('[size in MB]'), help="Specify a max request payload size. Any requests to the server larger than this size will be dropped. Do not change if unsure.", type=int, default=32)
advparser.add_argument("--overridekv", metavar=('[name=type:value]'), help="Advanced option to override a metadata by key, same as in llama.cpp. Mainly for debugging, not intended for general use. Types: int, float, bool, str", default="")
advparser.add_argument("--overridetensors", metavar=('[tensor name pattern=buffer type]'), help="Advanced option to override tensor backend selection, same as in llama.cpp.", default="")
advparser.add_argument("--overridekv","--override-kv", metavar=('[name=type:value]'), help="Advanced option to override a metadata by key, same as in llama.cpp. Mainly for debugging, not intended for general use. Types: int, float, bool, str", default="")
advparser.add_argument("--overridetensors","--override-tensor","-ot", metavar=('[tensor name pattern=buffer type]'), help="Advanced option to override tensor backend selection, same as in llama.cpp.", default="")
compatgroup2 = parser.add_mutually_exclusive_group()
compatgroup2.add_argument("--showgui", help="Always show the GUI instead of launching the model right away when loading settings from a .kcpps file.", action='store_true')
compatgroup2.add_argument("--skiplauncher", help="Doesn't display or use the GUI launcher. Overrides showgui.", action='store_true')
@ -7696,7 +7698,7 @@ if __name__ == '__main__':
deprecatedgroup.add_argument("--hordeconfig", help=argparse.SUPPRESS, nargs='+')
deprecatedgroup.add_argument("--sdconfig", help=argparse.SUPPRESS, nargs='+')
compatgroup.add_argument("--noblas", help=argparse.SUPPRESS, action='store_true')
compatgroup3.add_argument("--nommap", help=argparse.SUPPRESS, action='store_true')
compatgroup3.add_argument("--nommap","--no-mmap", help=argparse.SUPPRESS, action='store_true')
deprecatedgroup.add_argument("--sdnotile", help=argparse.SUPPRESS, action='store_true') # legacy option, see sdtiledvae
main(launch_args=parser.parse_args(),default_args=parser.parse_args([]))

2
otherarch/tts_adapter.cpp
View file

@ -779,7 +779,7 @@ static tts_generation_outputs ttstype_generate_ttscpp(const tts_generation_input
}
else if(detectedarch=="dia")
{
vmapper = {"zoe", "zac", "jess", "leo", "mia"};
vmapper = {"zac", "zoe", "jess", "leo", "mia"};
vpermitted = {"zoe", "zac","jess", "leo", "mia", "julia", "leah"};
}