Crate rlx_cli 

Shared pieces for per-model rlx-<family> binaries and the optional rlx-run multiplexer.

Re-exports

pub use args::req;

pub use auto_dispatch::SniffedFrom;

pub use auto_dispatch::SniffedRunner;

pub use auto_dispatch::UnimplementedArch;

pub use auto_dispatch::arch_runner_name;

pub use auto_dispatch::auto_dispatch;

pub use auto_dispatch::auto_runner_name;

pub use auto_dispatch::auto_sniff;

pub use auto_dispatch::known_unimplemented_arch;

pub use auto_dispatch::known_unimplemented_keys;

pub use auto_dispatch::model_type_runner_name;

pub use auto_dispatch::run_auto;

pub use chat::ChatMessage;

pub use chat::ChatTemplate;

pub use chat::ChatTemplateSource;

pub use chat::auto_chat_template;

pub use compat::CompatSource;

pub use compat::CompatibilityReport;

pub use compat::CompatibilityStatus;

pub use compat::GgufRequiredFields;

pub use compat::check_hf_repo;

pub use compat::check_path;

pub use compat::looks_like_hf_repo;

pub use compat::run_check;

pub use device::parse_device;

pub use device::parse_gemma_device;

pub use device::parse_llada2_device;

pub use device::parse_llama32_device;

pub use device::parse_qwen35_device;

pub use device::parse_sam_device;

pub use device::parse_standard_device;

pub use format::WeightFormat;

pub use inspect::estimate_qwen35_footprint;

pub use inspect::fmt_bytes;

pub use inspect::list_mtp_keys;

pub use inspect::run_inspect;

pub use lm_runner::LmRunner;

pub use loader::debug_resolve_name;

pub use loader::open_gguf_loader;

pub use loader::open_loader;

pub use loader::open_loader_resolved;

pub use loader::open_loader_resolved_with_options;

pub use loader::open_loader_with_format;

pub use loader::open_weight_map_resolved;

pub use loader::open_weights_resolved;

pub use mtmd::AssembledTurn;

pub use mtmd::MediaSource;

pub use mtmd::MtmdContext;

pub use mtmd::MtmdTurn;

pub use registry::ModelRunner;

pub use registry::dispatch;

pub use registry::dispatch_help;

pub use registry::register_cli;

pub use registry::register_runner;

pub use registry::registered_runners;

pub use registry::run_registered;

pub use weights_resolve::WeightsResolveCli;

pub use weights_resolve::resolve_weights_cli;

Modules

args

auto_dispatch

Auto-dispatch: pick a registered model runner from a weights path.

chat

Chat-template engine for RLX runners.

compat

Model-compatibility discovery for rlx-models.

device

format

inspect

lm_runner

Boxed-trait dispatch for LM runners (PLAN.md M3 + M8).

loader

mtmd

Multimodal turn assembly (PLAN.md M7).

registry

weights

Model-agnostic weight I/O — paths, formats, drain policy only.

weights_resolve

Shared --weights resolution for model CLIs (directory quants, split hints).

Structs

GgufDirGuide

RegisteredFormat

One registered on-disk format (built-in or custom).

WeightFormatRegistration

Describes one on-disk weight format.

Constants

STANDARD_DEVICE_NAMES

CLI / help string for --device.

Traits

GgufTensorNameResolver

Resolve a builder-requested tensor name to the name stored in a GGUF file.

Functions

gguf_dir_guide

Numbered .gguf listing + resolve hints for a directory (CLI / errors).

list_registered_formats

All registered formats (built-ins first, then custom registrations).

load_weight_map_resolved

Convenience: resolve + drain to F32 WeightMap.

load_weights_resolved

Resolve a file or directory, enforce GGUF arch policy, open via registry, optionally drain.

open_map

Resolve + drain to F32 WeightMap.

open_map_with

Resolve + drain with options.

open_weights

Resolve + open (live WeightLoader).

open_with

Resolve + open with options.

register_gguf_tensor_resolver

Register a custom resolver (call before first GGUF load). Later registrations win among resolvers that match the same architecture.

register_weight_format

Register a custom weight format (call before the first load). Later entries override built-ins when the same extension is registered twice.

validate_sam_device

SAM v1 also documents tpu on [rlx_sam::Sam::from_safetensors_on].

validate_standard_device

Fail fast on exotic runtime devices (TPU, ANE, OpenGL, …).

Type Aliases

LoadOpts

Alias for LoadWeightsOptions.

ResolveOpts

Alias for ResolveWeightsOptions.