Struct EmbeddingsConfig 

pub struct EmbeddingsConfig {

    pub model_repo: String,
    pub model_revision: String,
    pub model_file: String,
    pub tokenizer_file: String,
    pub pooling_config_file: String,
    pub transformer_config_file: String,
    pub max_length: usize,
    pub normalize: bool,
    pub intra_threads: Option<usize>,
    pub cache_dir: Option<PathBuf>,
    pub local_model_path: Option<PathBuf>,
    pub local_tokenizer_path: Option<PathBuf>,
    pub local_pooling_config_path: Option<PathBuf>,
    pub local_transformer_config_path: Option<PathBuf>,
    pub input_ids_name: Option<String>,
    pub attention_mask_name: Option<String>,
    pub token_type_ids_name: Option<String>,
    pub output_name: Option<String>,
}

Configuration for the built-in ONNX Runtime embedder.

Fields

model_repo: String

Hugging Face model repository used when local assets are not supplied.

model_revision: String

Hugging Face model revision used when local assets are not supplied.

model_file: String

Model file path inside the Hugging Face repository.

tokenizer_file: String

Tokenizer file path inside the Hugging Face repository.

pooling_config_file: String

Pooling config file path inside the Hugging Face repository.

transformer_config_file: String

Transformer config file path inside the Hugging Face repository.

max_length: usize

Maximum tokenizer sequence length, capped by model config when known.

normalize: bool

Whether output embeddings should be L2-normalized.

intra_threads: Option<usize>

Optional ONNX Runtime intra-op thread count.

cache_dir: Option<PathBuf>

Optional Hugging Face cache directory.

local_model_path: Option<PathBuf>

Local ONNX model path. If unset, the model is resolved from Hugging Face.

local_tokenizer_path: Option<PathBuf>

Local tokenizer path. If unset, the tokenizer is resolved from Hugging Face.

local_pooling_config_path: Option<PathBuf>

Local sentence-transformers pooling config path.

local_transformer_config_path: Option<PathBuf>

Local transformer config path.

input_ids_name: Option<String>

Override for models that use a non-standard input IDs name.

attention_mask_name: Option<String>

Override for models that use a non-standard attention mask name.

token_type_ids_name: Option<String>

Override for models that use a non-standard token type IDs name.

output_name: Option<String>

Optional output tensor name. Defaults to the first model output.

Trait Implementations

impl Clone for EmbeddingsConfig

fn clone(&self) -> EmbeddingsConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for EmbeddingsConfig

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

Formats the value using the given formatter.

impl Default for EmbeddingsConfig

fn default() -> Self

Returns the “default value” for a type.

impl PartialEq for EmbeddingsConfig

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Eq for EmbeddingsConfig

impl StructuralPartialEq for EmbeddingsConfig