fastembed 4.4.0

Rust implementation of https://github.com/qdrant/fastembed
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167

//! Initialization options for the text embedding models.
//!

use crate::{
    common::{TokenizerFiles, DEFAULT_CACHE_DIR},
    pooling::Pooling,
    EmbeddingModel, QuantizationMode,
};
use ort::{execution_providers::ExecutionProviderDispatch, session::Session};
use std::path::{Path, PathBuf};
use tokenizers::Tokenizer;

use super::{DEFAULT_EMBEDDING_MODEL, DEFAULT_MAX_LENGTH};

/// Options for initializing the TextEmbedding model
#[derive(Debug, Clone)]
#[non_exhaustive]
pub struct InitOptions {
    pub model_name: EmbeddingModel,
    pub execution_providers: Vec<ExecutionProviderDispatch>,
    pub max_length: usize,
    pub cache_dir: PathBuf,
    pub show_download_progress: bool,
}

impl InitOptions {
    /// Create a new InitOptions with the given model name
    pub fn new(model_name: EmbeddingModel) -> Self {
        Self {
            model_name,
            ..Default::default()
        }
    }

    /// Set the maximum length of the input text
    pub fn with_max_length(mut self, max_length: usize) -> Self {
        self.max_length = max_length;
        self
    }

    /// Set the cache directory for the model files
    pub fn with_cache_dir(mut self, cache_dir: PathBuf) -> Self {
        self.cache_dir = cache_dir;
        self
    }

    /// Set the execution providers for the model
    pub fn with_execution_providers(
        mut self,
        execution_providers: Vec<ExecutionProviderDispatch>,
    ) -> Self {
        self.execution_providers = execution_providers;
        self
    }

    /// Set whether to show download progress
    pub fn with_show_download_progress(mut self, show_download_progress: bool) -> Self {
        self.show_download_progress = show_download_progress;
        self
    }
}

impl Default for InitOptions {
    fn default() -> Self {
        Self {
            model_name: DEFAULT_EMBEDDING_MODEL,
            execution_providers: Default::default(),
            max_length: DEFAULT_MAX_LENGTH,
            cache_dir: Path::new(DEFAULT_CACHE_DIR).to_path_buf(),
            show_download_progress: true,
        }
    }
}

/// Options for initializing UserDefinedEmbeddingModel
///
/// Model files are held by the UserDefinedEmbeddingModel struct
#[derive(Debug, Clone)]
#[non_exhaustive]
pub struct InitOptionsUserDefined {
    pub execution_providers: Vec<ExecutionProviderDispatch>,
    pub max_length: usize,
}

impl InitOptionsUserDefined {
    pub fn new() -> Self {
        Self {
            ..Default::default()
        }
    }

    pub fn with_execution_providers(
        mut self,
        execution_providers: Vec<ExecutionProviderDispatch>,
    ) -> Self {
        self.execution_providers = execution_providers;
        self
    }

    pub fn with_max_length(mut self, max_length: usize) -> Self {
        self.max_length = max_length;
        self
    }
}

impl Default for InitOptionsUserDefined {
    fn default() -> Self {
        Self {
            execution_providers: Default::default(),
            max_length: DEFAULT_MAX_LENGTH,
        }
    }
}

/// Convert InitOptions to InitOptionsUserDefined
///
/// This is useful for when the user wants to use the same options for both the default and user-defined models
impl From<InitOptions> for InitOptionsUserDefined {
    fn from(options: InitOptions) -> Self {
        InitOptionsUserDefined {
            execution_providers: options.execution_providers,
            max_length: options.max_length,
        }
    }
}

/// Struct for "bring your own" embedding models
///
/// The onnx_file and tokenizer_files are expecting the files' bytes
#[derive(Debug, Clone, PartialEq, Eq)]
#[non_exhaustive]
pub struct UserDefinedEmbeddingModel {
    pub onnx_file: Vec<u8>,
    pub tokenizer_files: TokenizerFiles,
    pub pooling: Option<Pooling>,
    pub quantization: QuantizationMode,
}

impl UserDefinedEmbeddingModel {
    pub fn new(onnx_file: Vec<u8>, tokenizer_files: TokenizerFiles) -> Self {
        Self {
            onnx_file,
            tokenizer_files,
            quantization: QuantizationMode::None,
            pooling: None,
        }
    }

    pub fn with_quantization(mut self, quantization: QuantizationMode) -> Self {
        self.quantization = quantization;
        self
    }

    pub fn with_pooling(mut self, pooling: Pooling) -> Self {
        self.pooling = Some(pooling);
        self
    }
}

/// Rust representation of the TextEmbedding model
pub struct TextEmbedding {
    pub tokenizer: Tokenizer,
    pub(crate) pooling: Option<Pooling>,
    pub(crate) session: Session,
    pub(crate) need_token_type_ids: bool,
    pub(crate) quantization: QuantizationMode,
}