Add multilingual language and systems learning tracks

en/expanded-25-2/README.md

@@ -0,0 +1,9 @@
+# AI Runtime Engineering · 2
+
+[Agila University track](https://edu.agilaos.com/track.php?track=25&lang=en)
+
+About 15 minutes per lesson. Predict, run, change, explain, then complete the check in University.
+
+- [Lesson 1: Plan memory lifetimes](lesson-1/lesson.html)
+- [Lesson 2: Quantization has an error budget](lesson-2/lesson.html)
+- [Lesson 3: An explainable runtime capstone](lesson-3/lesson.html)

en/expanded-25-2/lesson-1/lesson.html

@@ -0,0 +1,4 @@
+<!doctype html><html lang="en"><meta charset="utf-8"><title>Plan memory lifetimes</title><h1>Plan memory lifetimes</h1><p>15-minute session: read for 3 minutes, trace for 3, experiment for 7, and complete the check for 2. Spend longer whenever you need.</p><p>An execution graph has values with creation and last-use points. Reuse a buffer only after its old value is no longer needed and no asynchronous operation is still using it. Weights, activations, temporary workspaces, and caches need separate accounting. Calculate element count times bytes per element, then include alignment and overhead in a real implementation.</p><p><strong>Practice and evidence:</strong> Predict the example or draw its state transitions. Change one assumption, test the boundary, and record the actual result. Explain what this small model does not establish about a production system.</p><pre><code class="language-python">shape=(4,8)
+bytes_per_element=4
+print(shape[0]*shape[1]*bytes_per_element)
+</code></pre><p><strong>Expected result:</strong> <code>128</code></p><h3>Practice and evidence</h3><p>Keep three notes: your prediction, the actual result, and why they match or differ. The completion check grades understanding; the practice task builds skill.</p><p><a href="https://git.agilaos.com/matt/open_agx">Language reference: git.agilaos.com</a></p><p><a href="https://onnx.ai/onnx/intro/">Language reference: onnx.ai</a></p></html>

en/expanded-25-2/lesson-1/main.py

@@ -0,0 +1,3 @@
+shape=(4,8)
+bytes_per_element=4
+print(shape[0]*shape[1]*bytes_per_element)

en/expanded-25-2/lesson-2/lesson.html

@@ -0,0 +1,8 @@
+<!doctype html><html lang="en"><meta charset="utf-8"><title>Quantization has an error budget</title><h1>Quantization has an error budget</h1><p>15-minute session: read for 3 minutes, trace for 3, experiment for 7, and complete the check for 2. Spend longer whenever you need.</p><p>A simple symmetric quantizer divides by a scale, rounds, and clamps to an integer range. Dequantization multiplies by the scale. Clipping and rounding lose information. Measure both numerical error and the task’s accuracy before calling a smaller representation acceptable. This scalar exercise is not an implementation of every production quantization format.</p><p><strong>Practice and evidence:</strong> Predict the example or draw its state transitions. Change one assumption, test the boundary, and record the actual result. Explain what this small model does not establish about a production system.</p><pre><code class="language-python">value=0.36
+scale=0.1
+q=max(-127,min(127,round(value/scale)))
+restored=q*scale
+print(q)
+print(round(abs(value-restored),3))
+</code></pre><p><strong>Expected result:</strong> <code>4
+0.04</code></p><h3>Practice and evidence</h3><p>Keep three notes: your prediction, the actual result, and why they match or differ. The completion check grades understanding; the practice task builds skill.</p><p><a href="https://git.agilaos.com/matt/open_agx">Language reference: git.agilaos.com</a></p><p><a href="https://onnx.ai/onnx/intro/">Language reference: onnx.ai</a></p></html>

en/expanded-25-2/lesson-2/main.py

@@ -0,0 +1,6 @@
+value=0.36
+scale=0.1
+q=max(-127,min(127,round(value/scale)))
+restored=q*scale
+print(q)
+print(round(abs(value-restored),3))

en/expanded-25-2/lesson-3/lesson.html

@@ -0,0 +1 @@
+<!doctype html><html lang="en"><meta charset="utf-8"><title>An explainable runtime capstone</title><h1>An explainable runtime capstone</h1><p>15-minute session: read for 3 minutes, trace for 3, experiment for 7, and complete the check for 2. Spend longer whenever you need.</p><p>Build a tiny graph executor with input, matrix multiplication, bias, and activation nodes. Validate shapes, keep a CPU reference, plan buffer lifetimes, and benchmark warm and cold runs separately. AGX Open currently provides a CPU reference backend; a requested GPU backend is not evidence of GPU execution. Add a backend only when detection and correctness checks prove it is used.</p><p><strong>Practice and evidence:</strong> Predict the example or draw its state transitions. Change one assumption, test the boundary, and record the actual result. Explain what this small model does not establish about a production system.</p><h3>Practice and evidence</h3><p>Keep three notes: your prediction, the actual result, and why they match or differ. The completion check grades understanding; the practice task builds skill.</p><p><a href="https://git.agilaos.com/matt/open_agx">Language reference: git.agilaos.com</a></p><p><a href="https://onnx.ai/onnx/intro/">Language reference: onnx.ai</a></p></html>