uteAI 0.1.0__tar.gz → 0.2.0__tar.gz

uteai-0.2.0/PKG-INFO

@@ -0,0 +1,60 @@
+Metadata-Version: 2.4
+Name: uteAI
+Version: 0.2.0
+Summary: Advanced AI, Neural Network, and Edge Engine Library from HCMUTE
+Author-email: Dung Cai <dung.cai@hcmute.edu.vn>
+License: MIT
+Project-URL: Homepage, https://github.com/your-username/uteAI
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Requires-Dist: numpy>=1.20.0
+
+# uteAI
+**Advanced AI, Neural Network, and Edge Engine Library from HCMUTE**
+
+`uteAI` is a high-performance, educational Deep Learning engine built from scratch. It implements standard feedforward neural networks with seamless CPU/GPU duality using NumPy and CuPy.
+
+## Key Features
+* **Seamless CPU/GPU Duality**: Automatically detects and utilizes NVIDIA CUDA GPUs via `CuPy`. Safely falls back to `NumPy` if no GPU is available.
+* **Zero-Copy Host-to-Device Pipeline**: Minimizes PCIe transfer bottlenecks during training.
+* **Built from Scratch**: Clear, educational implementations of Dense layers, Activations (ReLU, Softmax, Sigmoid), Optimizers (Adam, RMSprop, SGD), and Loss functions (Categorical Cross-Entropy, MSE).
+* **Cross-Platform Portability**: Model weights are saved as standard NumPy arrays, ensuring checkpoints are compatible across purely CPU or GPU environments.
+
+## Installation
+```bash
+pip install uteAI
+```
+
+## Quick Start
+```python
+import uteAI.core as nn
+import numpy as np
+
+# Automatically selects 'cuda' if available, else 'cpu'
+model = nn.Model(device='auto')
+
+# Build architecture
+model.add(nn.Dense(784, 128, weight_init='he'))
+model.add(nn.Activation_ReLU())
+model.add(nn.Dense_Dropout(0.1))
+model.add(nn.Dense(128, 10))
+model.add(nn.Activation_SoftMax())
+
+# Compile model
+model.set(
+    loss=nn.Loss_CategoricalCrossEntropy(),
+    optimizer=nn.Optimizer_Adam(learning_rate=0.001),
+    accuracy=nn.Accuracy_Categorical()
+)
+model.finalize()
+
+# Train (assuming X_train, y_train are prepared)
+# model.train(X_train, y_train, epochs=10, batch_size=128)
+```
+
+## Author
+Developed by **Dung Cai** - Ho Chi Minh City University of Technology and Education (HCMUTE).

uteai-0.2.0/README.md

@@ -0,0 +1,45 @@
+# uteAI
+**Advanced AI, Neural Network, and Edge Engine Library from HCMUTE**
+
+`uteAI` is a high-performance, educational Deep Learning engine built from scratch. It implements standard feedforward neural networks with seamless CPU/GPU duality using NumPy and CuPy.
+
+## Key Features
+* **Seamless CPU/GPU Duality**: Automatically detects and utilizes NVIDIA CUDA GPUs via `CuPy`. Safely falls back to `NumPy` if no GPU is available.
+* **Zero-Copy Host-to-Device Pipeline**: Minimizes PCIe transfer bottlenecks during training.
+* **Built from Scratch**: Clear, educational implementations of Dense layers, Activations (ReLU, Softmax, Sigmoid), Optimizers (Adam, RMSprop, SGD), and Loss functions (Categorical Cross-Entropy, MSE).
+* **Cross-Platform Portability**: Model weights are saved as standard NumPy arrays, ensuring checkpoints are compatible across purely CPU or GPU environments.
+
+## Installation
+```bash
+pip install uteAI
+```
+
+## Quick Start
+```python
+import uteAI.core as nn
+import numpy as np
+
+# Automatically selects 'cuda' if available, else 'cpu'
+model = nn.Model(device='auto')
+
+# Build architecture
+model.add(nn.Dense(784, 128, weight_init='he'))
+model.add(nn.Activation_ReLU())
+model.add(nn.Dense_Dropout(0.1))
+model.add(nn.Dense(128, 10))
+model.add(nn.Activation_SoftMax())
+
+# Compile model
+model.set(
+    loss=nn.Loss_CategoricalCrossEntropy(),
+    optimizer=nn.Optimizer_Adam(learning_rate=0.001),
+    accuracy=nn.Accuracy_Categorical()
+)
+model.finalize()
+
+# Train (assuming X_train, y_train are prepared)
+# model.train(X_train, y_train, epochs=10, batch_size=128)
+```
+
+## Author
+Developed by **Dung Cai** - Ho Chi Minh City University of Technology and Education (HCMUTE).

{uteai-0.1.0 → uteai-0.2.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "uteAI"
-version = "0.1.0"
+version = "0.2.0"
 description = "Advanced AI, Neural Network, and Edge Engine Library from HCMUTE"
 readme = "README.md"
 authors = [

uteai-0.2.0/uteAI.egg-info/PKG-INFO

@@ -0,0 +1,60 @@
+Metadata-Version: 2.4
+Name: uteAI
+Version: 0.2.0
+Summary: Advanced AI, Neural Network, and Edge Engine Library from HCMUTE
+Author-email: Dung Cai <dung.cai@hcmute.edu.vn>
+License: MIT
+Project-URL: Homepage, https://github.com/your-username/uteAI
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+Requires-Dist: numpy>=1.20.0
+
+# uteAI
+**Advanced AI, Neural Network, and Edge Engine Library from HCMUTE**
+
+`uteAI` is a high-performance, educational Deep Learning engine built from scratch. It implements standard feedforward neural networks with seamless CPU/GPU duality using NumPy and CuPy.
+
+## Key Features
+* **Seamless CPU/GPU Duality**: Automatically detects and utilizes NVIDIA CUDA GPUs via `CuPy`. Safely falls back to `NumPy` if no GPU is available.
+* **Zero-Copy Host-to-Device Pipeline**: Minimizes PCIe transfer bottlenecks during training.
+* **Built from Scratch**: Clear, educational implementations of Dense layers, Activations (ReLU, Softmax, Sigmoid), Optimizers (Adam, RMSprop, SGD), and Loss functions (Categorical Cross-Entropy, MSE).
+* **Cross-Platform Portability**: Model weights are saved as standard NumPy arrays, ensuring checkpoints are compatible across purely CPU or GPU environments.
+
+## Installation
+```bash
+pip install uteAI
+```
+
+## Quick Start
+```python
+import uteAI.core as nn
+import numpy as np
+
+# Automatically selects 'cuda' if available, else 'cpu'
+model = nn.Model(device='auto')
+
+# Build architecture
+model.add(nn.Dense(784, 128, weight_init='he'))
+model.add(nn.Activation_ReLU())
+model.add(nn.Dense_Dropout(0.1))
+model.add(nn.Dense(128, 10))
+model.add(nn.Activation_SoftMax())
+
+# Compile model
+model.set(
+    loss=nn.Loss_CategoricalCrossEntropy(),
+    optimizer=nn.Optimizer_Adam(learning_rate=0.001),
+    accuracy=nn.Accuracy_Categorical()
+)
+model.finalize()
+
+# Train (assuming X_train, y_train are prepared)
+# model.train(X_train, y_train, epochs=10, batch_size=128)
+```
+
+## Author
+Developed by **Dung Cai** - Ho Chi Minh City University of Technology and Education (HCMUTE).

{uteai-0.1.0 → uteai-0.2.0}/uteAI.egg-info/SOURCES.txt

@@ -1,3 +1,4 @@
+README.md
 pyproject.toml
 tests/test_core.py
 uteAI/__init__.py

uteai-0.1.0/PKG-INFO

@@ -1,14 +0,0 @@
-Metadata-Version: 2.4
-Name: uteAI
-Version: 0.1.0
-Summary: Advanced AI, Neural Network, and Edge Engine Library from HCMUTE
-Author-email: Dung Cai <dung.cai@hcmute.edu.vn>
-License: MIT
-Project-URL: Homepage, https://github.com/your-username/uteAI
-Classifier: Programming Language :: Python :: 3
-Classifier: License :: OSI Approved :: MIT License
-Classifier: Operating System :: OS Independent
-Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
-Requires-Python: >=3.8
-Description-Content-Type: text/markdown
-Requires-Dist: numpy>=1.20.0

uteai-0.1.0/uteAI.egg-info/PKG-INFO

@@ -1,14 +0,0 @@
-Metadata-Version: 2.4
-Name: uteAI
-Version: 0.1.0
-Summary: Advanced AI, Neural Network, and Edge Engine Library from HCMUTE
-Author-email: Dung Cai <dung.cai@hcmute.edu.vn>
-License: MIT
-Project-URL: Homepage, https://github.com/your-username/uteAI
-Classifier: Programming Language :: Python :: 3
-Classifier: License :: OSI Approved :: MIT License
-Classifier: Operating System :: OS Independent
-Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
-Requires-Python: >=3.8
-Description-Content-Type: text/markdown
-Requires-Dist: numpy>=1.20.0