v1.2

README.md

@@ -1,32 +1,37 @@
-# RustML
-Open-source AI Library for Rust <br />
-__v1.1__
-
-## About
-`RustML` is an AI/machine learning library built for the `Rust` programming language. It's desgined to work similarly to `Python` libraries, such as PyTorch & Tensorflow. 
-
-## Current Functions & Features
-- Sigmoid function
-```rust
-Sigmoid(x: f64) -> f64
-```
-
-- ReLU function
-```rust
-ReLU(x: f64) -> f64
-```
-
-- Softmax function
-```rust
-Softmax(x: &[f64]) -> f64
-```
-
-- RNN algoritm
-```rust
-RNN(x: &[f64], y: &[f64], weights: &[f64], bias: f64) -> Vec<f64> 
-```
-
-- LSTM algoritm
-```rust
-LSTM(x: &[f64], y: &[f64], weights: &[f64]) -> Vec<f64> 
-```
+# RustML
+Open-source AI Library for Rust <br />
+__v1.1__
+
+## About
+`RustML` is an AI/machine learning library built for the `Rust` programming language. It's desgined to work similarly to `Python` libraries, such as PyTorch & Tensorflow. 
+
+## Current Functions & Features
+- Sigmoid function
+```rust
+Sigmoid(x: f64) -> f64
+```
+
+- ReLU function
+```rust
+ReLU(x: f64) -> f64
+```
+
+- Softmax function
+```rust
+Softmax(x: &[f64]) -> f64
+```
+
+- RNN algoritm
+```rust
+RNN(x: &[f64], y: &[f64], weights: &[f64], bias: f64) -> Vec<f64> 
+```
+
+- LSTM algoritm
+```rust
+LSTM(x: &[f64], y: &[f64], weights: &[f64]) -> Vec<f64> 
+```
+
+- Gradient
+```rust
+gradient(x: f64, y: f64, loss: f64) -> f64
+```

src/rustml.rs

@@ -25,6 +25,10 @@ impl RustML {
         exp_x.iter().map(|&v| v as f64 / sum_exp_x as f64).sum()
     }
 
+    pub fn gradient(x: f64, y: f64, loss: f64) -> f64 {
+        2.0 * x * (loss - y)
+    }
+
     pub fn ReLU(x: f64) -> f64 {
         x.max(0.0)
     }