llmgw ctl guard judge status lists each judge model the policies ask: in the catalog, as a judge, replicas ready, its last failure to answer, and what is wrong in words. enable serves the missing one through model serve - Llama Guard 3 1B on CPUs, role guard, under the name the policies give it - with --replace for a model of that name that is not a judge; the console has it under Do. GET /api/v1/guard/judges, and the summary's judges, put the problem on ctl status and the console's Overview. A Llama Guard file served from the mirror page is added as a judge, under the name the policies ask, rather than as a model clients use the policies never find. ctl says plainly when the API answers with a page. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
84 lines
5.0 KiB
JSON
84 lines
5.0 KiB
JSON
{
|
||
"serve.refusal": "Cannot be served as it stands: %s.",
|
||
"serve.serving": "Serving <code>%s</code> with <strong>%s</strong>.",
|
||
"serve.whole_version": "The whole version, %d files: the node fetches them into one directory and the engine is given the directory.",
|
||
"serve.name": "Name",
|
||
"serve.aliases": "Aliases",
|
||
"serve.aliases_hint": "comma separated",
|
||
"serve.aliases_example": "coder-7b",
|
||
"serve.replicas": "Replicas",
|
||
"serve.runs": "Runs",
|
||
"serve.mode.ondemand": "on demand: loads when asked, stops after 15 min unused",
|
||
"serve.description": "What it is for",
|
||
"serve.description_hint": "shown to clients in /v1/models",
|
||
"serve.description_example": "general chat, strong Swedish; 14B, slower than the 8B",
|
||
"serve.context_length": "Context a request may use",
|
||
"serve.max_output": "Output limit",
|
||
"serve.optional": "optional",
|
||
"serve.ctx_pool": "Engine context pool",
|
||
"serve.ctx_pool_hint": "divided across slots",
|
||
"serve.placement": "Runs on",
|
||
"serve.place.gpu": "GPU: %s, %d of %d free",
|
||
"serve.place.cpu_slower": "CPU only: no GPU, many times slower",
|
||
"serve.place.cpu_only": "CPU: this fleet has no GPU",
|
||
"serve.gpus": "GPUs per replica",
|
||
"serve.gpus_hint": "on one node",
|
||
"serve.parallel": "Parallel requests",
|
||
"serve.parallel_hint": "per replica",
|
||
"serve.cpu_ok": "Run on the CPU anyway: %d GPUs are free and would sit unused",
|
||
"serve.memory": "Needs RAM (GB)",
|
||
"serve.memory_hint": "for the context pool beside it",
|
||
"serve.cpu_cores": "CPU cores",
|
||
"serve.cpu_cores_hint": "reserved; the engine's threads",
|
||
"serve.selector": "Only nodes labelled",
|
||
"serve.selector_hint": "key / value, optional",
|
||
"serve.selector_key_example": "role",
|
||
"serve.selector_value_example": "cpu-large",
|
||
"serve.add": "Add to the catalog",
|
||
|
||
"serve.on_cpu": "%s on the CPU",
|
||
"serve.on_gpu_one": "%s on 1 GPU %s",
|
||
"serve.on_gpus": "%s on %d %s GPUs of one node",
|
||
"serve.host_memory": "%s GB of host memory",
|
||
"serve.cores.one": "%d core",
|
||
"serve.cores.other": "%d cores",
|
||
"serve.pool_parallel": "context pool %s over %s parallel requests (%s each)",
|
||
"serve.pool": "context pool %s",
|
||
"serve.will_run_as": "Will run as: %s.",
|
||
"serve.free_gpus_warning": "%s GPUs are free; on the CPU this runs many times slower.",
|
||
|
||
"serve.pool_note": "The context pool is what the engine is given; llama.cpp divides it across its parallel slots, so the number one request may use is the smaller one above.",
|
||
"serve.memory_note": "Memory: %s. Change the context pool and this changes with it, by the KV cache's share — an estimate, and editable. It is filled in rather than left blank because a CPU profile has no card to be short of: with no figure, every ready node is a candidate and the first one picked is whichever the scheduler reached, not whichever can hold the model. A node label does the same job by name if you would rather pin it.",
|
||
"serve.basis.estimate": "weights plus half again plus two",
|
||
"serve.basis.shape": "weights %s GB + KV cache %s GB for %s tokens (%d layers × %d KV heads) + %s GB for the engine",
|
||
|
||
"serve.refuse.no_engine": "nothing here reads %s files",
|
||
"serve.refuse.llamacpp": "llama.cpp on the nodes cannot run this: %s",
|
||
"serve.refuse.declared": "a model called %s is declared in configuration; edit it there rather than adding a second",
|
||
"serve.warn.not_installed": "no node reports having %s installed, and %s weights need it",
|
||
"serve.warn.no_gpu": "This fleet has no GPU either; a GGUF build of the same model runs on the CPU nodes",
|
||
"serve.warn.too_big": "the largest node has %d GB and this needs about %d",
|
||
"serve.warn.and": "%s, and %s",
|
||
|
||
"serve.err.name_source": "a model needs a name and a source",
|
||
"serve.err.whole_number": "%s must be a whole number",
|
||
"serve.what.context_length": "the context length",
|
||
"serve.what.max_output": "the output limit",
|
||
"serve.what.replicas": "the replica count",
|
||
"serve.what.gpus": "the GPUs per replica",
|
||
"serve.what.ctx_pool": "the engine's context pool",
|
||
"serve.what.parallel": "the parallel requests",
|
||
"serve.what.memory": "the memory requirement",
|
||
"serve.what.cores": "the cores",
|
||
"serve.err.engine": "which engine should run it?",
|
||
"serve.err.needs_gpu": "%s runs on GPUs; choose a GPU class to place it on",
|
||
"serve.err.placement": "%q is not a placement",
|
||
"serve.err.cpu_with_free_gpus.one": "this fleet has %d GPU free and %s was set to run on the CPU; choose a GPU placement, or tick \"run on the CPU anyway\"",
|
||
"serve.err.cpu_with_free_gpus.other": "this fleet has %d GPUs free and %s was set to run on the CPU; choose a GPU placement, or tick \"run on the CPU anyway\"",
|
||
"serve.msg.added": "%s added to the catalog, %s: %s; it is placed on the next pass",
|
||
"serve.replicas_n.one": "%d replica",
|
||
"serve.replicas_n.other": "%d replicas",
|
||
"serve.msg.judge": "It is the guardrails' judge (role: guard): policies naming %s ask it; clients do not see it.",
|
||
"serve.msg.judge_renamed": "Its weights are a judge's, so it is the guardrails' judge (role: guard), named %[2]s as the policies ask rather than %[1]s."
|
||
}
|