merge master to branch (#2146)

* Set a security policy (#2142) * Add security policy Signed-off-by: Pedro Kaj Kjellerup Nacht <[email protected]> * Only use private vuln report Signed-off-by: Pedro Kaj Kjellerup Nacht <[email protected]> --------- Signed-off-by: Pedro Kaj Kjellerup Nacht <[email protected]> * add Random ops shim (#2145) * add Random ops shim * update random opss * handle None case * change seed handeling * undo changes * undo changes * undo * update int_seed * change int to init * update init_seed everywhere * add kwargs * undo bial * code reformat * remove dtype --------- Signed-off-by: Pedro Kaj Kjellerup Nacht <[email protected]> Co-authored-by: Pedro Kaj Kjellerup Nacht <[email protected]>
keras-team · Nov 14, 2023 · 3385708 · 3385708
1 parent e52523a
commit 3385708
Show file tree

Hide file tree

Showing 2 changed files with 137 additions and 1 deletion.
diff --git a/SECURITY.md b/SECURITY.md
@@ -0,0 +1,18 @@
+# Security Policy
+
+If you have discovered a security vulnerability in this project, please report it
+privately. **Do not disclose it as a public issue.** This gives us time to work with you
+to fix the issue before public exposure, reducing the chance that the exploit will be
+used before a patch is released.
+
+You may submit the report in the following ways:
+
+- send a [private vulnerability report](https://github.com/keras-team/keras-cv/security/advisories/new)
+
+Please provide the following information in your report:
+
+- A description of the vulnerability and its impact
+- How to reproduce the issue
+
+This project is maintained by volunteers on a reasonable-effort basis. As such,
+please give us 90 days to work on a fix before public exposure.
diff --git a/keras_cv/backend/random.py b/keras_cv/backend/random.py
@@ -11,10 +11,128 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
+from keras_cv.backend import keras
 from keras_cv.backend.config import keras_3
 
 if keras_3():
     from keras.random import *  # noqa: F403, F401
 else:
     from keras_core.random import *  # noqa: F403, F401
+
+
+class SeedGenerator:
+    def __init__(self, seed=None, **kwargs):
+        if keras_3():
+            self._seed_generator = keras.random.SeedGenerator(
+                seed=seed, **kwargs
+            )
+        else:
+            self._current_seed = [0, seed]
+
+    def next(self, ordered=True):
+        if keras_3():
+            return self._seed_generator.next(ordered=ordered)
+        else:
+            self._current_seed[0] += 1
+            return self._current_seed[:]
+
+
+def normal(shape, mean=0.0, stddev=1.0, dtype=None, seed=None):
+    if isinstance(seed, SeedGenerator):
+        seed = seed.next()
+        init_seed = seed[0] + seed[1]
+    else:
+        init_seed = seed
+
+    kwargs = {}
+    if dtype:
+        kwargs["dtype"] = dtype
+    if keras_3():
+        return keras.random.normal(
+            shape,
+            mean=mean,
+            stddev=stddev,
+            seed=init_seed,
+            **kwargs,
+        )
+    else:
+        import tensorflow as tf
+
+        return tf.random.normal(
+            shape,
+            mean=mean,
+            stddev=stddev,
+            seed=init_seed,
+            **kwargs,
+        )
+
+
+def uniform(shape, minval=0.0, maxval=1.0, dtype=None, seed=None):
+    if isinstance(seed, SeedGenerator):
+        seed = seed.next()
+        init_seed = seed[0] + seed[1]
+    else:
+        init_seed = seed
+    kwargs = {}
+    if dtype:
+        kwargs["dtype"] = dtype
+    if keras_3():
+        return keras.random.uniform(
+            shape,
+            minval=minval,
+            maxval=maxval,
+            seed=init_seed,
+            **kwargs,
+        )
+    else:
+        import tensorflow as tf
+
+        return tf.random.uniform(
+            shape,
+            minval=minval,
+            maxval=maxval,
+            seed=init_seed,
+            **kwargs,
+        )
+
+
+def shuffle(x, axis=0, seed=None):
+    if isinstance(seed, SeedGenerator):
+        seed = seed.next()
+        init_seed = seed[0] + seed[1]
+    else:
+        init_seed = seed
+
+    if keras_3():
+        return keras.random.shuffle(x=x, axis=axis, seed=init_seed)
+    else:
+        import tensorflow as tf
+
+        return tf.random.shuffle(x=x, axis=axis, seed=init_seed)
+
+
+def categorical(logits, num_samples, dtype=None, seed=None):
+    if isinstance(seed, SeedGenerator):
+        seed = seed.next()
+        init_seed = seed[0] + seed[1]
+    else:
+        init_seed = seed
+    kwargs = {}
+    if dtype:
+        kwargs["dtype"] = dtype
+    if keras_3():
+        return keras.random.categorical(
+            logits=logits,
+            num_samples=num_samples,
+            seed=init_seed,
+            **kwargs,
+        )
+    else:
+        import tensorflow as tf
+
+        return tf.random.categorical(
+            logits=logits,
+            num_samples=num_samples,
+            seed=init_seed,
+            **kwargs,
+        )