grx-tensor 0.2.0 → 0.2.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
metadata CHANGED
@@ -1,7 +1,7 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: grx-tensor
3
3
  version: !ruby/object:Gem::Version
4
- version: 0.2.0
4
+ version: 0.2.1
5
5
  platform: ruby
6
6
  authors:
7
7
  - Razo
@@ -67,8 +67,9 @@ dependencies:
67
67
  version: '2.0'
68
68
  description: |
69
69
  GRX brings PyTorch-style tensor operations to Ruby. Every arithmetic op,
70
- activation, and optimizer step runs through a native C library compiled
71
- with AVX2+FMA SIMD. Ruby is the interface C does the work.
70
+ activation, and optimizer step runs through a native C library with
71
+ dynamic multi-target SIMD dispatch (AVX2+FMA, SSE, and scalar fallback).
72
+ Ruby is the interface — C does the work.
72
73
 
73
74
  Features: autograd, SGD/Adam optimizers, Linear/Sequential/Dropout/BatchNorm
74
75
  layers, MSE/BCE/CrossEntropy loss functions, Xavier and He weight init.
@@ -112,36 +113,39 @@ metadata:
112
113
  bug_tracker_uri: https://github.com/Gabo-Razo/grx-tensor/issues
113
114
  post_install_message: |2+
114
115
 
115
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
116
- GRX-Tensor 0.2.0 installed
117
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
116
+ ============================================================
117
+ GRX-Tensor 0.2.1
118
+ ============================================================
118
119
 
119
- Compile the C extension to enable AVX2+FMA SIMD:
120
+ [ENGLISH]
121
+ Thank you for choosing and using GRX-Tensor!
120
122
 
121
- Linux / macOS: make -C ext/unix
122
- Windows: make -C ext/windows -f Makefile.mingw
123
+ * Native C acceleration with dynamic SIMD dispatch
124
+ (AVX2+FMA, SSE, and portable scalar C) is built and
125
+ configured automatically.
126
+ * Windows Note: Native C acceleration is supported when
127
+ using RubyInstaller with DevKit (MSYS2 / MinGW-w64).
128
+ Standalone pre-compiled Windows binaries are currently
129
+ in active development.
130
+ * Documentation, guides, and tutorials:
131
+ https://github.com/Gabo-Razo/grx-tensor
123
132
 
124
- Without it, GRX runs in pure Ruby fallback mode (slower but correct).
133
+ ----------------------------------------------------------
125
134
 
126
- Quick start:
135
+ [ESPAÑOL]
136
+ Muchas gracias por elegir y utilizar GRX-Tensor!
127
137
 
128
- require "grx"
138
+ * La aceleracion nativa en C con despacho dinamico SIMD
139
+ (AVX2+FMA, SSE y C escalar portable) se compila y
140
+ configura de forma totalmente automatica.
141
+ * Nota para Windows: La aceleracion nativa esta soportada
142
+ al utilizar RubyInstaller con DevKit (MSYS2 / MinGW-w64).
143
+ Los binarios pre-compilados independientes para Windows
144
+ se encuentran actualmente en desarrollo activo.
145
+ * Documentacion, guias y tutoriales:
146
+ https://github.com/Gabo-Razo/grx-tensor
129
147
 
130
- a = GRX.tensor([1.0, 2.0, 3.0], [3], requires_grad: true)
131
- b = GRX.tensor([4.0, 5.0, 6.0], [3], requires_grad: true)
132
- c = a + b
133
- c.backward
134
- puts a.grad.to_a # [1.0, 1.0, 1.0]
135
-
136
- net = GRX::NN::Sequential.new(
137
- GRX::NN::Linear.new(4, 16),
138
- GRX::NN::ReLU.new,
139
- GRX::NN::Linear.new(16, 1)
140
- )
141
- opt = GRX::Optim::Adam.new(net.parameters, lr: 0.001)
142
-
143
- Docs: https://github.com/Gabo-Razo/grx-tensor
144
- ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
148
+ ============================================================
145
149
 
146
150
  rdoc_options: []
147
151
  require_paths: