Update CI TF matrix and add example smoke coverage

Author: shenweichen

.github/workflows/ci.yml

@@ -1,74 +1,84 @@
 name: CI
 
-on: 
+on:
   push:
-    path:
-      - 'ge/*'
-      - 'tests/*'
+    paths:
+      - "ge/**"
+      - "tests/**"
+      - "examples/**"
+      - ".github/workflows/**"
+      - "setup.py"
+      - "README.md"
   pull_request:
-    path:
-      - 'ge/*'
-      - 'tests/*'
-      
+    paths:
+      - "ge/**"
+      - "tests/**"
+      - "examples/**"
+      - ".github/workflows/**"
+      - "setup.py"
+      - "README.md"
+
 jobs:
   build:
-
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-22.04
     timeout-minutes: 180
     strategy:
+      fail-fast: false
       matrix:
-        python-version: [3.6,3.7,3.8]
-        tf-version: [1.4.0,1.15.0,2.5.0,2.6.0,2.7.0,2.8.0,2.9.0]
-        
-        exclude:
-          - python-version: 3.7
-            tf-version: 1.4.0
-          - python-version: 3.7
-            tf-version: 1.15.0
-          - python-version: 3.8
-            tf-version: 1.4.0
-          - python-version: 3.8
-            tf-version: 1.14.0
-          - python-version: 3.8
-            tf-version: 1.15.0
-          - python-version: 3.6
-            tf-version: 2.7.0
-          - python-version: 3.6
-            tf-version: 2.8.0
-          - python-version: 3.6
-            tf-version: 2.9.0
-          - python-version: 3.9
-            tf-version: 1.4.0
-          - python-version: 3.9
-            tf-version: 1.15.0
-          - python-version: 3.9
-            tf-version: 2.2.0
+        include:
+          - python-version: "3.7"
+            tf-version: "1.15.5"
+            use-legacy-keras: "0"
+          - python-version: "3.10"
+            tf-version: "2.10.0"
+            use-legacy-keras: "0"
+          - python-version: "3.10"
+            tf-version: "2.15.0"
+            use-legacy-keras: "0"
+          - python-version: "3.11"
+            tf-version: "2.15.0"
+            use-legacy-keras: "0"
+          - python-version: "3.10"
+            tf-version: "2.20.0"
+            use-legacy-keras: "1"
+          - python-version: "3.11"
+            tf-version: "2.20.0"
+            use-legacy-keras: "1"
+          - python-version: "3.12"
+            tf-version: "2.20.0"
+            use-legacy-keras: "0"
+
     steps:
-    
-    - uses: actions/checkout@v3
-    
-    - name: Setup python environment
-      uses: actions/setup-python@v4
-      with:
-        python-version: ${{ matrix.python-version }}
+      - uses: actions/checkout@v4
+
+      - name: Setup Python environment
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install -q "tensorflow==${{ matrix.tf-version }}"
+          if [[ "${{ matrix.tf-version }}" == 1.* ]]; then
+            python -m pip install -q "protobuf==3.20.3"
+          fi
+          if [[ "${{ matrix.use-legacy-keras }}" == "1" ]]; then
+            python -m pip install -q "tf-keras~=2.20"
+          fi
+          python -m pip install -e ".[test]"
+
+      - name: Test with pytest
+        timeout-minutes: 180
+        env:
+          TF_USE_LEGACY_KERAS: ${{ matrix.use-legacy-keras }}
+        run: |
+          pytest --cov=ge --cov=examples --cov-report=xml
 
-    - name: Install dependencies
-      run: |
-        pip3 install -q tensorflow==${{ matrix.tf-version }}
-        pip install -q protobuf==3.19.0
-        pip install -q requests
-        pip install -e .
-    - name: Test with pytest
-      timeout-minutes: 180
-      run: |
-        pip install -q pytest
-        pip install -q pytest-cov
-        pip install -q python-coveralls
-        pytest --cov=ge --cov-report=xml
-    - name: Upload coverage to Codecov  
-      uses: codecov/codecov-action@v3.1.0
-      with:
-        token: ${{secrets.CODECOV_TOKEN}}
-        file: ./coverage.xml
-        flags: pytest
-        name: py${{ matrix.python-version }}-tf${{ matrix.tf-version }}
+      - name: Upload coverage to Codecov
+        uses: codecov/codecov-action@v4
+        with:
+          token: ${{ secrets.CODECOV_TOKEN }}
+          file: ./coverage.xml
+          flags: pytest
+          name: py${{ matrix.python-version }}-tf${{ matrix.tf-version }}

README.md

@@ -21,15 +21,34 @@
 | Struc2Vec | [KDD 2017][struc2vec: Learning Node Representations from Structural Identity](https://arxiv.org/pdf/1704.03165.pdf)        | [【Graph Embedding】Struc2Vec：算法原理，实现和应用](https://zhuanlan.zhihu.com/p/56733145) |
 
 
+# CI Compatibility Matrix
+
+The CI matrix currently validates the following Python/TensorFlow combinations:
+
+| Python | TensorFlow | `TF_USE_LEGACY_KERAS` |
+| :----: | :--------: | :-------------------: |
+| 3.7    | 1.15.5     | 0 |
+| 3.10   | 2.10.0     | 0 |
+| 3.10   | 2.15.0     | 0 |
+| 3.11   | 2.15.0     | 0 |
+| 3.10   | 2.20.0     | 1 |
+| 3.11   | 2.20.0     | 1 |
+| 3.12   | 2.20.0     | 0 |
+
+For TensorFlow 2.16+ jobs that need legacy Keras behavior, CI installs `tf-keras` and sets `TF_USE_LEGACY_KERAS=1`.
+
 # How to run examples
-1. clone the repo and make sure you have installed `tensorflow` or `tensorflow-gpu` on your local machine. 
-2. run following commands
+
+1. Clone the repo and install dependencies.
+2. Run one example script.
+
 ```bash
-python setup.py install
-cd examples
-python deepwalk_wiki.py
+pip install -e .[cpu]
+python examples/deepwalk_wiki.py
 ```
 
+Each example now exposes `main(smoke=False, show=True)`. CI executes all `examples/*.py` in smoke mode (`smoke=True`, `show=False`) to keep runtime short while still validating the training/import paths.
+
 ## DisscussionGroup & Related Projects
 
 <html>

examples/alias.py

@@ -4,27 +4,37 @@
 from ge.alias import alias_sample, create_alias_table
 
 
-def gen_prob_dist(N):
-    p = np.random.randint(0, 100, N)
-    return p/np.sum(p)
+def gen_prob_dist(size):
+    probabilities = np.random.randint(0, 100, size)
+    return probabilities / np.sum(probabilities)
 
 
-def simulate(N=100, k=10000,):
+def simulate(size=100, sample_count=10000):
+    truth = gen_prob_dist(size)
+    accept, alias = create_alias_table(truth)
 
-    truth = gen_prob_dist(N)
+    sampled = np.zeros(size)
+    for _ in range(sample_count):
+        sampled[alias_sample(accept, alias)] += 1
+    return sampled / np.sum(sampled), truth
 
-    area_ratio = truth
-    accept, alias = create_alias_table(area_ratio)
 
-    ans = np.zeros(N)
-    for _ in range(k):
-        i = alias_sample(accept, alias)
-        ans[i] += 1
-    return ans/np.sum(ans), truth
+def main(smoke=False, show=True):
+    size = 20 if smoke else 100
+    sample_count = 300 if smoke else 10000
+    alias_result, truth = simulate(size=size, sample_count=sample_count)
+
+    assert np.isclose(alias_result.sum(), 1.0)
+    assert np.isclose(truth.sum(), 1.0)
+
+    if show:
+        plt.bar(list(range(len(alias_result))), alias_result, label="alias_result")
+        plt.bar(list(range(len(truth))), truth, label="truth")
+        plt.legend()
+        plt.show()
+
+    return alias_result, truth
 
 
 if __name__ == "__main__":
-    alias_result, truth = simulate()
-    plt.bar(list(range(len(alias_result))), alias_result, label='alias_result')
-    plt.bar(list(range(len(truth))), truth, label='truth')
-    plt.legend()
+    main()

examples/deepwalk_wiki.py

@@ -1,53 +1,73 @@
+from pathlib import Path
 
+import matplotlib.pyplot as plt
+import networkx as nx
 import numpy as np
+from sklearn.linear_model import LogisticRegression
+from sklearn.manifold import TSNE
 
-from ge.classify import read_node_label, Classifier
 from ge import DeepWalk
-from sklearn.linear_model import LogisticRegression
+from ge.classify import Classifier, read_node_label
 
-import matplotlib.pyplot as plt
-import networkx as nx
-from sklearn.manifold import TSNE
+PROJECT_ROOT = Path(__file__).resolve().parents[1]
+WIKI_GRAPH_PATH = PROJECT_ROOT / "data" / "wiki" / "Wiki_edgelist.txt"
+WIKI_LABEL_PATH = PROJECT_ROOT / "data" / "wiki" / "wiki_labels.txt"
+SMOKE_GRAPH_PATH = PROJECT_ROOT / "tests" / "Wiki_edgelist.txt"
 
 
-def evaluate_embeddings(embeddings):
-    X, Y = read_node_label('../data/wiki/wiki_labels.txt')
-    tr_frac = 0.8
-    print("Training classifier using {:.2f}% nodes...".format(
-        tr_frac * 100))
+def evaluate_embeddings(embeddings, label_path):
+    x_data, y_data = read_node_label(str(label_path))
+    train_fraction = 0.8
+    print("Training classifier using {:.2f}% nodes...".format(train_fraction * 100))
     clf = Classifier(embeddings=embeddings, clf=LogisticRegression())
-    clf.split_train_evaluate(X, Y, tr_frac)
-
+    clf.split_train_evaluate(x_data, y_data, train_fraction)
 
-def plot_embeddings(embeddings,):
-    X, Y = read_node_label('../data/wiki/wiki_labels.txt')
 
-    emb_list = []
-    for k in X:
-        emb_list.append(embeddings[k])
-    emb_list = np.array(emb_list)
+def plot_embeddings(embeddings, label_path, show=True):
+    x_data, y_data = read_node_label(str(label_path))
 
-    model = TSNE(n_components=2)
-    node_pos = model.fit_transform(emb_list)
+    embedding_list = np.array([embeddings[node] for node in x_data])
+    node_pos = TSNE(n_components=2).fit_transform(embedding_list)
 
     color_idx = {}
-    for i in range(len(X)):
-        color_idx.setdefault(Y[i][0], [])
-        color_idx[Y[i][0]].append(i)
+    for index, label in enumerate(y_data):
+        color_idx.setdefault(label[0], [])
+        color_idx[label[0]].append(index)
 
-    for c, idx in color_idx.items():
-        plt.scatter(node_pos[idx, 0], node_pos[idx, 1], label=c)
+    for label, indexes in color_idx.items():
+        plt.scatter(node_pos[indexes, 0], node_pos[indexes, 1], label=label)
     plt.legend()
-    plt.show()
+    if show:
+        plt.show()
+    else:
+        plt.close()
 
 
-if __name__ == "__main__":
-    G = nx.read_edgelist('../data/wiki/Wiki_edgelist.txt',
-                         create_using=nx.DiGraph(), nodetype=None, data=[('weight', int)])
+def main(smoke=False, show=True):
+    graph_path = SMOKE_GRAPH_PATH if smoke else WIKI_GRAPH_PATH
+    graph = nx.read_edgelist(
+        str(graph_path),
+        create_using=nx.DiGraph(),
+        nodetype=None,
+        data=[("weight", int)],
+    )
 
-    model = DeepWalk(G, walk_length=10, num_walks=80, workers=1)
-    model.train(window_size=5, iter=3)
+    model = DeepWalk(
+        graph,
+        walk_length=3 if smoke else 10,
+        num_walks=2 if smoke else 80,
+        workers=1,
+    )
+    model.train(window_size=2 if smoke else 5, iter=1 if smoke else 3, workers=1)
     embeddings = model.get_embeddings()
+    assert len(embeddings) > 0
+
+    if not smoke:
+        evaluate_embeddings(embeddings, WIKI_LABEL_PATH)
+        plot_embeddings(embeddings, WIKI_LABEL_PATH, show=show)
 
-    evaluate_embeddings(embeddings)
-    plot_embeddings(embeddings)
+    return embeddings
+
+
+if __name__ == "__main__":
+    main()