Correct spelling and documentation errors across the repository

README.md

@@ -90,7 +90,7 @@ Quantized versions are also available for several models above (append `Q` to th
 - [**BAAI/bge-reranker-base**](https://huggingface.co/BAAI/bge-reranker-base) - Default
 - [**BAAI/bge-reranker-v2-m3**](https://huggingface.co/BAAI/bge-reranker-v2-m3)
 - [**jinaai/jina-reranker-v1-turbo-en**](https://huggingface.co/jinaai/jina-reranker-v1-turbo-en)
-- [**jinaai/jina-reranker-v2-base-multiligual**](https://huggingface.co/jinaai/jina-reranker-v2-base-multilingual)
+- [**jinaai/jina-reranker-v2-base-multilingual**](https://huggingface.co/jinaai/jina-reranker-v2-base-multilingual)
 
 </details>
 

src/bgem3_embedding/init.rs

@@ -26,7 +26,7 @@ pub struct Bgem3EmbeddingOutput {
 
 /// Struct for "bring your own" BGE-M3 models
 ///
-/// The onnx_file and tokenizer_files expect the files' bytes
+/// The `onnx_file` and `tokenizer_files` fields contain the file contents as bytes.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct UserDefinedBgem3Model {
     pub onnx_file: Vec<u8>,

src/common.rs

@@ -107,8 +107,8 @@ pub struct TokenizerFiles {
     pub tokenizer_config_file: Vec<u8>,
 }
 
-/// The procedure for loading tokenizer files from the hugging face hub is separated
-/// from the main load_tokenizer function (which is expecting bytes, from any source).
+/// Loading tokenizer files from the Hugging Face Hub is separate from the main
+/// `load_tokenizer` function, which expects bytes from any source.
 #[cfg(feature = "hf-hub")]
 pub fn load_tokenizer_hf_hub(model_repo: ApiRepo, max_length: usize) -> Result<Tokenizer> {
     let tokenizer_files: TokenizerFiles = TokenizerFiles {
@@ -159,7 +159,8 @@ pub fn load_tokenizer(tokenizer_files: TokenizerFiles, max_length: usize) -> Res
             )
         })?;
 
-    //For BGEBaseSmall, the model_max_length value is set to 1000000000000000019884624838656. Which fits in a f64
+    // For BGEBaseSmall, model_max_length is set to 1000000000000000019884624838656.
+    // This value fits in an f64.
     let model_max_length = tokenizer_config["model_max_length"]
         .as_f64()
         .ok_or_else(|| {

src/image_embedding/init.rs

@@ -70,7 +70,7 @@ impl From<ImageInitOptions> for ImageInitOptionsUserDefined {
 
 /// Struct for "bring your own" embedding models
 ///
-/// The onnx_file and preprocessor_files are expecting the files' bytes
+/// The `onnx_file` and `preprocessor_file` fields contain the file contents as bytes.
 #[derive(Debug, Clone, PartialEq, Eq)]
 #[non_exhaustive]
 pub struct UserDefinedImageEmbeddingModel {

src/image_embedding/utils.rs

@@ -144,7 +144,7 @@ impl Transform for CenterCrop {
                 Array3::zeros((3usize, crop_height as usize, crop_width as usize));
             let offset_x = (crop_width - origin_width) / 2;
             let offset_y = (crop_height - origin_height) / 2;
-            // whc -> chw
+            // HWC -> CHW
             for (x, y, pixel) in image.to_rgb8().enumerate_pixels() {
                 pixels_array[[0, (y + offset_y) as usize, (x + offset_x) as usize]] =
                     pixel[0] as f32;
@@ -166,7 +166,7 @@ impl Transform for PILToNDarray {
             TransformData::Image(image) => {
                 let image = image.to_rgb8();
                 let (width, height) = image.dimensions();
-                // whc -> chw
+                // HWC -> CHW
                 let mut pixels_array = Array3::zeros((3usize, height as usize, width as usize));
                 for (x, y, pixel) in image.enumerate_pixels() {
                     pixels_array[[0, y as usize, x as usize]] = pixel[0] as f32;

src/models/image_embedding.rs

@@ -40,7 +40,7 @@ pub fn models_list() -> Vec<ModelInfo<ImageEmbeddingModel>> {
         ModelInfo {
             model: ImageEmbeddingModel::UnicomVitB16,
             dim: 768,
-            description: String::from("Unicom Unicom-ViT-B-16 from open-metric-learning"),
+            description: String::from("Unicom-ViT-B-16 from Open Metric Learning"),
             model_code: String::from("Qdrant/Unicom-ViT-B-16"),
             model_file: String::from("model.onnx"),
             additional_files: Vec::new(),
@@ -49,7 +49,7 @@ pub fn models_list() -> Vec<ModelInfo<ImageEmbeddingModel>> {
         ModelInfo {
             model: ImageEmbeddingModel::UnicomVitB32,
             dim: 512,
-            description: String::from("Unicom Unicom-ViT-B-32 from open-metric-learning"),
+            description: String::from("Unicom-ViT-B-32 from Open Metric Learning"),
             model_code: String::from("Qdrant/Unicom-ViT-B-32"),
             model_file: String::from("model.onnx"),
             additional_files: Vec::new(),
@@ -58,7 +58,7 @@ pub fn models_list() -> Vec<ModelInfo<ImageEmbeddingModel>> {
         ModelInfo {
             model: ImageEmbeddingModel::NomicEmbedVisionV15,
             dim: 768,
-            description: String::from("Nomic NomicEmbedVisionV15"),
+            description: String::from("Nomic Embed Vision v1.5"),
             model_code: String::from("nomic-ai/nomic-embed-vision-v1.5"),
             model_file: String::from("onnx/model.onnx"),
             additional_files: Vec::new(),

src/models/reranking.rs

@@ -12,7 +12,13 @@ pub enum RerankerModel {
     /// jinaai/jina-reranker-v1-turbo-en
     JINARerankerV1TurboEn,
     /// jinaai/jina-reranker-v2-base-multilingual
-    JINARerankerV2BaseMultiligual,
+    JINARerankerV2BaseMultilingual,
+}
+
+impl RerankerModel {
+    #[allow(non_upper_case_globals)]
+    #[deprecated(note = "use `JINARerankerV2BaseMultilingual` instead")]
+    pub const JINARerankerV2BaseMultiligual: Self = Self::JINARerankerV2BaseMultilingual;
 }
 
 pub fn reranker_model_list() -> Vec<RerankerModelInfo> {
@@ -26,7 +32,7 @@ pub fn reranker_model_list() -> Vec<RerankerModelInfo> {
         },
         RerankerModelInfo {
             model: RerankerModel::BGERerankerV2M3,
-            description: String::from("reranker model for multilingual"),
+            description: String::from("Multilingual reranker model"),
             model_code: String::from("rozgo/bge-reranker-v2-m3"),
             model_file: String::from("model.onnx"),
             additional_files: vec![String::from("model.onnx.data")],
@@ -39,8 +45,8 @@ pub fn reranker_model_list() -> Vec<RerankerModelInfo> {
             additional_files: vec![],
         },
         RerankerModelInfo {
-            model: RerankerModel::JINARerankerV2BaseMultiligual,
-            description: String::from("reranker model for multilingual"),
+            model: RerankerModel::JINARerankerV2BaseMultilingual,
+            description: String::from("Multilingual reranker model"),
             model_code: String::from("jinaai/jina-reranker-v2-base-multilingual"),
             model_file: String::from("onnx/model.onnx"),
             additional_files: vec![],
@@ -86,14 +92,14 @@ pub(crate) fn all_variants() -> Vec<RerankerModel> {
             RerankerModel::BGERerankerBase => (),
             RerankerModel::BGERerankerV2M3 => (),
             RerankerModel::JINARerankerV1TurboEn => (),
-            RerankerModel::JINARerankerV2BaseMultiligual => (),
+            RerankerModel::JINARerankerV2BaseMultilingual => (),
         }
     }
     vec![
         RerankerModel::BGERerankerBase,
         RerankerModel::BGERerankerV2M3,
         RerankerModel::JINARerankerV1TurboEn,
-        RerankerModel::JINARerankerV2BaseMultiligual,
+        RerankerModel::JINARerankerV2BaseMultilingual,
     ]
 }
 

src/models/sparse.rs

@@ -31,7 +31,7 @@ pub fn models_list() -> Vec<ModelInfo<SparseModel>> {
             model: SparseModel::BGEM3,
             dim: 0,
             description: String::from(
-                "BGE-M3 sparse embedding model with 8192 context, supports 100+ languages",
+                "BGE-M3 sparse embedding model with an 8192-token context, supporting 100+ languages",
             ),
             model_code: String::from("BAAI/bge-m3"),
             model_file: String::from("onnx/model.onnx"),

src/reranking/init.rs

@@ -21,9 +21,9 @@ impl HasMaxLength for RerankerModel {
 /// Options for initializing the reranking models
 pub type RerankInitOptions = InitOptionsWithLength<RerankerModel>;
 
-/// Options for initializing UserDefinedRerankerModel
+/// Options for initializing a user-defined reranking model.
 ///
-/// Model files are held by the UserDefinedRerankerModel struct
+/// Model files are held by the [`UserDefinedRerankingModel`] struct.
 #[derive(Debug, Clone)]
 #[non_exhaustive]
 pub struct RerankInitOptionsUserDefined {
@@ -144,7 +144,7 @@ impl From<PathBuf> for OnnxSource {
 
 /// Struct for "bring your own" reranking models
 ///
-/// The onnx_file and tokenizer_files are expecting the files' bytes
+/// The `onnx_file` and `tokenizer_files` fields contain the file contents as bytes.
 #[derive(Debug, Clone, PartialEq, Eq)]
 #[non_exhaustive]
 pub struct UserDefinedRerankingModel {
@@ -174,7 +174,7 @@ mod tests {
     use super::*;
 
     #[test]
-    fn userdefined_builders_set_fields() {
+    fn user_defined_builders_set_fields() {
         let o = RerankInitOptionsUserDefined::new()
             .with_max_length(128)
             .with_intra_threads(2)

src/sparse_text_embedding/init.rs

@@ -19,7 +19,7 @@ pub type SparseInitOptions = InitOptionsWithLength<SparseModel>;
 
 /// Struct for "bring your own" sparse embedding models
 ///
-/// The onnx_file and tokenizer_files are expecting the files' bytes
+/// The `onnx_file` and `tokenizer_files` fields contain the file contents as bytes.
 #[derive(Debug, Clone, PartialEq, Eq)]
 #[non_exhaustive]
 pub struct UserDefinedSparseModel {

src/text_embedding/impl.rs

@@ -63,7 +63,7 @@ impl TextEmbedding {
             })?;
         }
 
-        // prioritise loading pooling config if available, if not (thanks qdrant!), look for it in hardcoded
+        // Prioritize loading the pooling config if available; otherwise, look for it in the hardcoded defaults.
         let post_processing = TextEmbedding::get_default_pooling_method(&model_name);
 
         let session = init_session_builder(execution_providers, intra_threads, session_config)?

src/text_embedding/init.rs

@@ -22,7 +22,7 @@ impl HasMaxLength for EmbeddingModel {
 /// Options for initializing the TextEmbedding model
 pub type TextInitOptions = InitOptionsWithLength<EmbeddingModel>;
 
-/// Options for initializing UserDefinedEmbeddingModel
+/// Options for initializing a user-defined embedding model.
 ///
 /// Model files are held by the UserDefinedEmbeddingModel struct
 #[derive(Debug, Clone)]
@@ -150,7 +150,7 @@ impl From<TextInitOptions> for InitOptionsUserDefined {
 
 /// Struct for "bring your own" embedding models
 ///
-/// The onnx_file and tokenizer_files are expecting the files' bytes
+/// The `onnx_file` and `tokenizer_files` fields contain the file contents as bytes.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct UserDefinedEmbeddingModel {
     pub onnx_file: Vec<u8>,
@@ -163,8 +163,8 @@ pub struct UserDefinedEmbeddingModel {
 
 /// Struct for adding external initializers to "bring your own" embedding models
 ///
-/// The buffer is expecting the data of the external initializer and the file_name
-/// must match the one referenced by the model.
+/// The buffer contains the external initializer's data, and `file_name` must
+/// match the name referenced by the model.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct ExternalInitializerFile {
     pub file_name: String,

tests/text-embeddings.rs

@@ -38,7 +38,7 @@ fn verify_embeddings(model: &EmbeddingModel, embeddings: &[Embedding]) -> Result
         EmbeddingModel::AllMiniLML12V2Q => [-0.07808663, 0.27919534, -0.0770612, -0.75660324],
         EmbeddingModel::AllMiniLML6V2 => [0.59605527, 0.36542925, -0.16450031, -0.40903988],
         EmbeddingModel::AllMiniLML6V2Q => [0.5677276, 0.40180072, -0.15454668, -0.4672576],
-        EmbeddingModel::AllMpnetBaseV2=> [-0.21253541, -0.050802127, 0.14072442, -0.2908188],
+        EmbeddingModel::AllMpnetBaseV2 => [-0.21253541, -0.050802127, 0.14072442, -0.2908188],
         EmbeddingModel::BGEBaseENV15 => [-0.51290065, -0.4844747, -0.53036124, -0.5337459],
         EmbeddingModel::BGEBaseENV15Q => [-0.5130697, -0.48461288, -0.53067875, -0.5337806],
         EmbeddingModel::BGELargeENV15 => [-0.19347441, -0.28394595, -0.1549195, -0.22201893],
@@ -547,7 +547,7 @@ fn test_bgesmallen1point5_match_python_counterpart() {
     // Tokenize with python SentenceTransformer("BAAI/bge-small-en-v1.5") default tokenizer
     // with (text, padding="max_length",max_length=384,truncation=True, return_tensors="np").
     // Normalized and pooled with SentenceTransformer("BAAI/bge-small-en-v1.5") default pooling settings.
-    // we only take a 10 items to keep the test file polite
+    // We only take 10 items to keep the test file small.
     let baseline: Vec<f32> = vec![
         4.208_193_7e-2,
         -2.748_133_2e-2,
@@ -586,7 +586,7 @@ fn test_allminilml6v2_match_python_counterpart() {
     // Tokenizer with python SentenceTransformer("all-mini-lm-l6-v2") default tokenizer
     // with (text, padding="max_length",max_length=384,truncation=True, return_tensors="np").
     // Normalized and pooled with SentenceTransformer("all-mini-lm-l6-v2") default pooling settings.
-    // we only take a 10 items to keep the test file polite
+    // We only take 10 items to keep the test file small.
     let baseline: Vec<f32> = vec![
         3.510_517_6e-2,
         1.046_043e-2,