Enum ModelFormat 

pub enum ModelFormat {
    Safetensors,
    GGUF,
    PyTorch,
    TensorRT,
    ONNX,
    MLX,
    CoreML,
    TorchScript,
    TFLite,
    TensorFlow,
    Keras,
    OpenVINO,
    TVM,
    NCNN,
    MNN,
    RKNN,
    Caffe,
    MXNet,
    Darknet,
    HDF5,
    Pickle,
    NumPy,
    Custom(String),
}

Supported AI model formats

Variants

Safetensors

Safetensors (.safetensors) - HuggingFace default for Transformers

GGUF

GGUF (.gguf) - Quantized LLM format (llama.cpp, LM Studio, Ollama)

PyTorch

PyTorch weights (.pt, .pth, .bin) - Classic state_dict files

TensorRT

TensorRT Engine (.plan) - NVIDIA compiled engines

ONNX

ONNX (.onnx) - Interchange/serving format

MLX

MLX (.npz) - Apple Silicon optimized format

CoreML

Core ML (.mlmodel) - iOS/macOS on-device inference

TorchScript

TorchScript (.pt traced/scripted) - PyTorch serialization

TFLite

TensorFlow Lite (.tflite) - Mobile/edge deployment

TensorFlow

TensorFlow SavedModel - TensorFlow serving format

Keras

Keras (.h5, .keras) - Keras model format

OpenVINO

OpenVINO IR (.xml + .bin) - Intel optimization format

TVM

TVM compiled artifacts

NCNN

NCNN (.param + .bin) - Mobile-optimized format

MNN

MNN (.mnn) - Mobile Neural Network format

RKNN

RKNN (.rknn) - Rockchip NPU format

Caffe

Caffe (.caffemodel) - Legacy Caffe format

MXNet

MXNet (.params) - Apache MXNet format

Darknet

Darknet (.weights) - YOLO/Darknet format

HDF5

HDF5 (.h5, .hdf5) - Hierarchical data format

Pickle

Pickle (.pkl) - Python pickle format

NumPy

NumPy (.npy, .npz) - NumPy array format

Custom(String)

Custom/Unknown format

Implementations

impl ModelFormat

pub fn from_extension(ext: &str) -> Self

Detect format from file extension

pub fn extension(&self) -> &str

Get file extension for format

pub fn from_name(name: &str) -> Option<Self>

Parse a format from its display name, case- and space-insensitively ("Core ML", "coreml", "CORE ML" all match).

pub fn from_stored(value: &str) -> Self

Parse a format string that may be either a display name or a file extension.

Version records persist format.name() (e.g. "PyTorch"), which from_extension does not recognise — it would silently yield Custom("pytorch") and break conversion-path lookup and tensor-level diffing. Use this whenever the input came from storage.

pub fn name(&self) -> &str

Get format name

pub fn telemetry_name(&self) -> &'static str

Format label safe to report in telemetry.

name returns the caller’s own string for ModelFormat::Custom, which is free-form user input — a model name, a path fragment, anything they typed. Reporting it would break the guarantee that telemetry carries no user-supplied text, so this collapses every Custom to the literal "custom" and is otherwise a fixed set of &'static str.

Trait Implementations

impl Clone for ModelFormat

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ModelFormat

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for ModelFormat

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Display for ModelFormat

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

Formats the value using the given formatter.

impl Eq for ModelFormat

impl Hash for ModelFormat

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for ModelFormat

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for ModelFormat

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for ModelFormat