diff --git a/README.md b/README.md
index d730313a66..5644d81813 100644
--- a/README.md
+++ b/README.md
@@ -112,7 +112,7 @@ We also provide a some command line based examples using state of the art models
 
 <img src="https://github.com/huggingface/candle/raw/main/candle-examples/examples/segment-anything/assets/sam_merged.jpg" width="200">
 
-- [SegFormer](./candle-examples/examples/segformer/): transformer based semantic segmantation model.
+- [SegFormer](./candle-examples/examples/segformer/): transformer based semantic segmentation model.
 - [Whisper](./candle-examples/examples/whisper/): speech recognition model.
 - [EnCodec](./candle-examples/examples/encodec/): high-quality audio compression
   model using residual vector quantization.
diff --git a/candle-book/src/lib.rs b/candle-book/src/lib.rs
index f8ca510ded..faffd7537b 100644
--- a/candle-book/src/lib.rs
+++ b/candle-book/src/lib.rs
@@ -81,7 +81,7 @@ let mut tp_shape = view.shape().to_vec();
 let size = tp_shape[0];
 
 if size % world_size != 0 {
-    panic!("The dimension is not divisble by `world_size`");
+    panic!("The dimension is not divisible by `world_size`");
 }
 let block_size = size / world_size;
 let start = rank * block_size;
diff --git a/candle-examples/src/bs1770.rs b/candle-examples/src/bs1770.rs
index e18b77a750..fbda6dfdc7 100644
--- a/candle-examples/src/bs1770.rs
+++ b/candle-examples/src/bs1770.rs
@@ -448,9 +448,9 @@ pub fn reduce_stereo_in_place(left: Windows100ms<&mut [Power]>, right: Windows10
 /// Perform gating and averaging for a BS.1770-4 integrated loudness measurement.
 ///
 /// The integrated loudness measurement is not just the average power over the
-/// entire signal. BS.1770-4 defines defines two stages of gating that exclude
+/// entire signal. BS.1770-4 defines two stages of gating that exclude
 /// parts of the signal, to ensure that silent parts do not contribute to the
-/// loudness measurment. This function performs that gating, and returns the
+/// loudness measurement. This function performs that gating, and returns the
 /// average power over the windows that were not excluded.
 ///
 /// The result of this function is the integrated loudness measurement.
diff --git a/candle-wasm-tests/tests/quantized_tests.rs b/candle-wasm-tests/tests/quantized_tests.rs
index fc107e611d..8705df4219 100644
--- a/candle-wasm-tests/tests/quantized_tests.rs
+++ b/candle-wasm-tests/tests/quantized_tests.rs
@@ -55,7 +55,7 @@ fn quantized_matmul_neg() -> Result<()> {
     Ok(())
 }
 
-/// Creates a vector simillarly to the one used in GGML unit tests: https://github.com/ggerganov/llama.cpp/blob/master/tests/test-quantize-fns.cpp#L26-L30
+/// Creates a vector similarly to the one used in GGML unit tests: https://github.com/ggerganov/llama.cpp/blob/master/tests/test-quantize-fns.cpp#L26-L30
 fn create_ggml_like_vector(offset: f32) -> Vec<f32> {
     const GGML_TEST_SIZE: usize = 32 * 128;
     (0..GGML_TEST_SIZE)