Enum LlmAction 

pub enum LlmAction {
    List {
        vault_path: Option<PathBuf>,
    },
    Duplicates {
        vault_path: Option<PathBuf>,
    },
    Load {
        model: String,
        vault_path: Option<PathBuf>,
    },
    Status,
    Eval {
        prompt: String,
        orchestrated: bool,
        stream: bool,
        lora: Option<PathBuf>,
    },
    Evict {
        model_id: String,
    },
    Test {
        vault_path: Option<PathBuf>,
        models: Option<Vec<String>>,
        quantization: Option<String>,
        verbose: bool,
    },
    Validate {
        vault_path: Option<PathBuf>,
        strict: bool,
    },
    ComprehensiveTest {
        vault_path: Option<PathBuf>,
        model: String,
        verbose: bool,
    },
    Benchmark {
        vault_path: Option<PathBuf>,
        models: Option<Vec<String>>,
        iterations: Option<u32>,
        warmup: Option<u32>,
    },
    Report {
        vault_path: Option<PathBuf>,
        output: Option<PathBuf>,
        format: Option<String>,
    },
    Convert {
        input: PathBuf,
        out: PathBuf,
        page_log2: u16,
        layout: String,
    },
    Optimize {
        input: PathBuf,
        out: Option<PathBuf>,
        skip_passport: bool,
    },
    Passport {
        reprobe: bool,
        gemv_n: usize,
        cache: Option<PathBuf>,
        apply_env_hint: bool,
        decode_proxy: Option<Option<PathBuf>>,
        decode_proxy_tokens: u32,
    },
    DecodeProxy {
        model: PathBuf,
        tokens: u32,
    },
    RawDecodeBench {
        model: PathBuf,
        steps: u32,
        warmups: u16,
        runs: u16,
        quantization: String,
        prompt: String,
        target_prompt_tokens: Option<u32>,
        retain_artifacts: Option<PathBuf>,
    },
    Mode {
        name: Option<String>,
    },
    PathSelect {
        reprobe: bool,
        apply: bool,
    },
    Profile {
        name: Option<String>,
    },
    Lab {
        action: String,
        model: Option<PathBuf>,
        tokens: u32,
        n_in: usize,
        n_out: usize,
        gemv_n: usize,
        out: Option<PathBuf>,
        hours: f64,
        max_generations: u32,
        ollama_model: Option<String>,
        ollama_url: String,
        no_ollama: bool,
    },
    Ground {
        prompt: String,
        answer: String,
    },
    SeedGrounding,
    CudaTcBench {
        side: usize,
    },
    Explore {
        input: PathBuf,
        out: Option<PathBuf>,
        tokens: u32,
        layouts: String,
        skip_convert: bool,
        sweep_ffn_f16: bool,
        modes: Option<String>,
    },
}

Variants

List

Fields

vault_path: Option<PathBuf>

Duplicates

Fields

vault_path: Option<PathBuf>

Load

Fields

model: String

vault_path: Option<PathBuf>

Status

Eval

Fields

prompt: String

orchestrated: bool

stream: bool

lora: Option<PathBuf>

Evict

Fields

model_id: String

Test

Fields

vault_path: Option<PathBuf>

models: Option<Vec<String>>

quantization: Option<String>

verbose: bool

Validate

Fields

vault_path: Option<PathBuf>

strict: bool

ComprehensiveTest

Fields

vault_path: Option<PathBuf>

model: String

verbose: bool

Benchmark

Fields

vault_path: Option<PathBuf>

models: Option<Vec<String>>

iterations: Option<u32>

warmup: Option<u32>

Report

Fields

vault_path: Option<PathBuf>

output: Option<PathBuf>

format: Option<String>

Convert

Fields

input: PathBuf

out: PathBuf

page_log2: u16

layout: String

Optimize

Fields

input: PathBuf

out: Option<PathBuf>

skip_passport: bool

Passport

Fields

reprobe: bool

gemv_n: usize

cache: Option<PathBuf>

apply_env_hint: bool

decode_proxy: Option<Option<PathBuf>>

decode_proxy_tokens: u32

DecodeProxy

Fields

model: PathBuf

tokens: u32

RawDecodeBench

Fixed-step model-only decoder benchmark (no product helpers, EOS, or fallback).

Fields

model: PathBuf

steps: u32

warmups: u16

runs: u16

quantization: String

prompt: String

target_prompt_tokens: Option<u32>

Deterministically cycle the encoded prompt to exactly this many prefill tokens.

retain_artifacts: Option<PathBuf>

Explicit evidence directory to create. It must not already exist.

Mode

Fields

name: Option<String>

PathSelect

Fields

reprobe: bool

apply: bool

Profile

Fields

name: Option<String>

Lab

Fields

action: String

model: Option<PathBuf>

tokens: u32

n_in: usize

n_out: usize

gemv_n: usize

out: Option<PathBuf>

hours: f64

max_generations: u32

ollama_model: Option<String>

ollama_url: String

no_ollama: bool

Ground

Fields

prompt: String

answer: String

SeedGrounding

CudaTcBench

Fields

side: usize

Explore

Fields

input: PathBuf

out: Option<PathBuf>

tokens: u32

layouts: String

skip_convert: bool

sweep_ffn_f16: bool

modes: Option<String>

Trait Implementations

impl Debug for LlmAction

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl FromArgMatches for LlmAction

fn from_arg_matches(__clap_arg_matches: &ArgMatches) -> Result<Self, Error>

Instantiate Self from [ArgMatches], parsing the arguments as needed.

fn from_arg_matches_mut( __clap_arg_matches: &mut ArgMatches, ) -> Result<Self, Error>

Instantiate Self from [ArgMatches], parsing the arguments as needed.

fn update_from_arg_matches( &mut self, __clap_arg_matches: &ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

fn update_from_arg_matches_mut<'b>( &mut self, __clap_arg_matches: &mut ArgMatches, ) -> Result<(), Error>

Assign values from ArgMatches to self.

impl Subcommand for LlmAction

fn augment_subcommands<'b>(__clap_app: Command) -> Command

Append to [Command] so it can instantiate Self via [FromArgMatches::from_arg_matches_mut]

fn augment_subcommands_for_update<'b>(__clap_app: Command) -> Command

Append to [Command] so it can instantiate self via [FromArgMatches::update_from_arg_matches_mut]

fn has_subcommand(__clap_name: &str) -> bool

Test whether Self can parse a specific subcommand