Pass directly the buffer ownership. (#949)

candle-wasm-examples/bert/src/bin/m.rs

@@ -18,8 +18,7 @@ impl Model {
         console_error_panic_hook::set_once();
         console_log!("loading model");
         let device = &Device::Cpu;
-        let weights = safetensors::tensor::SafeTensors::deserialize(&weights)?;
+        let vb = VarBuilder::from_buffered_safetensors(weights, DType::F64, device)?;
-        let vb = VarBuilder::from_safetensors(vec![weights], DType::F64, device);
         let config: Config = serde_json::from_slice(&config)?;
         let tokenizer =
             Tokenizer::from_bytes(&tokenizer).map_err(|m| JsError::new(&m.to_string()))?;

candle-wasm-examples/t5/src/bin/m.rs

@@ -29,8 +29,7 @@ impl ModelConditionalGeneration {
         console_error_panic_hook::set_once();
         console_log!("loading model");
         let device = &Device::Cpu;
-        let weights = safetensors::tensor::SafeTensors::deserialize(&weights)?;
+        let vb = VarBuilder::from_buffered_safetensors(weights, DType::F32, device)?;
-        let vb = VarBuilder::from_safetensors(vec![weights], DType::F32, device);
         let mut config: Config = serde_json::from_slice(&config)?;
         let tokenizer =
             Tokenizer::from_bytes(&tokenizer).map_err(|m| JsError::new(&m.to_string()))?;
@@ -128,8 +127,7 @@ impl ModelEncoder {
         console_error_panic_hook::set_once();
         console_log!("loading model");
         let device = &Device::Cpu;
-        let weights = safetensors::tensor::SafeTensors::deserialize(&weights)?;
+        let vb = VarBuilder::from_buffered_safetensors(weights, DType::F32, device)?;
-        let vb = VarBuilder::from_safetensors(vec![weights], DType::F32, device);
         let mut config: Config = serde_json::from_slice(&config)?;
         config.use_cache = false;
         let tokenizer =

candle-wasm-examples/whisper/src/worker.rs

@@ -253,8 +253,7 @@ impl Decoder {
         let mel_filters = mel_filters.tensor("mel_80")?.load(&device)?;
         console_log!("loaded mel filters {:?}", mel_filters.shape());
         let mel_filters = mel_filters.flatten_all()?.to_vec1::<f32>()?;
-        let weights = safetensors::tensor::SafeTensors::deserialize(&md.weights)?;
+        let vb = VarBuilder::from_buffered_safetensors(md.weights, DTYPE, &device)?;
-        let vb = VarBuilder::from_safetensors(vec![weights], DTYPE, &device);
         let config = Config::tiny_en();
         let whisper = Whisper::load(&vb, config)?;
         console_log!("done loading model");

candle-wasm-examples/yolo/src/bin/m.rs

@@ -13,7 +13,7 @@ pub struct Model {
 impl Model {
     #[wasm_bindgen(constructor)]
     pub fn new(data: Vec<u8>, model_size: &str) -> Result<Model, JsError> {
-        let inner = M::load_(&data, model_size)?;
+        let inner = M::load_(data, model_size)?;
         Ok(Self { inner })
     }
 
@@ -46,7 +46,7 @@ pub struct ModelPose {
 impl ModelPose {
     #[wasm_bindgen(constructor)]
     pub fn new(data: Vec<u8>, model_size: &str) -> Result<ModelPose, JsError> {
-        let inner = P::load_(&data, model_size)?;
+        let inner = P::load_(data, model_size)?;
         Ok(Self { inner })
     }
 

candle-wasm-examples/yolo/src/worker.rs

@@ -92,7 +92,7 @@ impl Model {
         Ok(bboxes)
     }
 
-    pub fn load_(weights: &[u8], model_size: &str) -> Result<Self> {
+    pub fn load_(weights: Vec<u8>, model_size: &str) -> Result<Self> {
         let multiples = match model_size {
             "n" => Multiples::n(),
             "s" => Multiples::s(),
@@ -104,14 +104,13 @@ impl Model {
             ))?,
         };
         let dev = &Device::Cpu;
-        let weights = safetensors::tensor::SafeTensors::deserialize(weights)?;
+        let vb = VarBuilder::from_buffered_safetensors(weights, DType::F32, dev)?;
-        let vb = VarBuilder::from_safetensors(vec![weights], DType::F32, dev);
         let model = YoloV8::load(vb, multiples, 80)?;
         Ok(Self { model })
     }
 
     pub fn load(md: ModelData) -> Result<Self> {
-        Self::load_(&md.weights, &md.model_size.to_string())
+        Self::load_(md.weights, &md.model_size.to_string())
     }
 }
 
@@ -172,7 +171,7 @@ impl ModelPose {
         Ok(bboxes)
     }
 
-    pub fn load_(weights: &[u8], model_size: &str) -> Result<Self> {
+    pub fn load_(weights: Vec<u8>, model_size: &str) -> Result<Self> {
         let multiples = match model_size {
             "n" => Multiples::n(),
             "s" => Multiples::s(),
@@ -184,14 +183,13 @@ impl ModelPose {
             ))?,
         };
         let dev = &Device::Cpu;
-        let weights = safetensors::tensor::SafeTensors::deserialize(weights)?;
+        let vb = VarBuilder::from_buffered_safetensors(weights, DType::F32, dev)?;
-        let vb = VarBuilder::from_safetensors(vec![weights], DType::F32, dev);
         let model = YoloV8Pose::load(vb, multiples, 1, (17, 3))?;
         Ok(Self { model })
     }
 
     pub fn load(md: ModelData) -> Result<Self> {
-        Self::load_(&md.weights, &md.model_size.to_string())
+        Self::load_(md.weights, &md.model_size.to_string())
     }
 }