tfds-nightly 4.9.8.dev202504110044__py3-none-any.whl → 4.9.8.dev202504130103__py3-none-any.whl

tensorflow_datasets/datasets/lbpp/CITATIONS.bib

@@ -0,0 +1,24 @@
+@inproceedings{matton-etal-2024-leakage,
+    title = "On Leakage of Code Generation Evaluation Datasets",
+    author = "Matton, Alexandre  and
+      Sherborne, Tom  and
+      Aumiller, Dennis  and
+      Tommasone, Elena  and
+      Alizadeh, Milad  and
+      He, Jingyi  and
+      Ma, Raymond  and
+      Voisin, Maxime  and
+      Gilsenan-McMahon, Ellen  and
+      Gall{\'e}, Matthias",
+    editor = "Al-Onaizan, Yaser  and
+      Bansal, Mohit  and
+      Chen, Yun-Nung",
+    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2024",
+    month = nov,
+    year = "2024",
+    address = "Miami, Florida, USA",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/2024.findings-emnlp.772/",
+    doi = "10.18653/v1/2024.findings-emnlp.772",
+    pages = "13215--13223",
+}

tensorflow_datasets/datasets/lbpp/README.md

@@ -0,0 +1,6 @@
+*Less Basic Python Programming* is a collection of 161 programming problems
+with accompanying unit tests.
+They were created with the aim of being fresh (not leaked at the time of
+creation) and more difficult than similar datasets (e.g., HumanEval and MBPP).
+It can serve as a drop-in replacement or enrichment of those datasets as they
+are structured in an equivalent way.

tensorflow_datasets/datasets/lbpp/__init__.py

@@ -0,0 +1,15 @@
+# coding=utf-8
+# Copyright 2024 The TensorFlow Datasets Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+

tensorflow_datasets/datasets/lbpp/lbpp_dataset_builder.py

@@ -0,0 +1,174 @@
+# coding=utf-8
+# Copyright 2024 The TensorFlow Datasets Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+
+# Copyright 2024 Cohere and the current dataset script contributor.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+# Author Note: Data loader is heavily inspired by the builder in
+# https://github.com/google-research/google-research/tree/main/lbpp_dataset
+"""Cohere Less Basic Python Problems. All columns decoded."""
+
+import base64
+import json
+import pickle
+import zlib
+
+from tensorflow_datasets.core.utils.lazy_imports_utils import pandas as pd
+import tensorflow_datasets.public_api as tfds
+
+
+_HOMEPAGE = "https://aclanthology.org/2024.findings-emnlp.772/"
+
+_VERSION = tfds.core.Version("2.0.0")
+
+_COLUMNS = [
+    "task_id",
+    "language",
+    "title",
+    "instruction",
+    "completion",
+    "test_file",
+    "test_list",
+    "signature",
+    "categories",
+    "test_setup",
+]
+
+_LANGUAGES = ["python", "cpp", "go", "java", "js", "rust"]
+_ALL_LANGUAGE_ALIASES = ["all", "multilingual"]
+_LANGUAGE_ALIAS_MAP = {
+    "default": "python",
+    "javascript": "js",
+}
+
+
+def decode_str(str_to_decode: str):
+  return json.loads(
+      pickle.loads(
+          zlib.decompress(base64.b64decode(str_to_decode.encode("utf-8")))
+      )
+  )
+
+
+class LBPPConfig(tfds.core.BuilderConfig):
+  """BuilderConfig."""
+
+  def __init__(self, name, description, features, **kwargs):
+    super(LBPPConfig, self).__init__(name=name, version=_VERSION, **kwargs)
+    self.name = name
+    self.description = description
+    self.features = features
+
+
+class Builder(tfds.core.GeneratorBasedBuilder):
+  """Builder for LBPP dataset."""
+
+  VERSION = _VERSION
+  LICENSE = "apache-2.0"
+  BUILDER_CONFIGS = [
+      LBPPConfig(
+          name="all", description="Multilingual LBPP", features=_COLUMNS
+      ),
+      LBPPConfig(
+          name="multilingual",
+          description="Multilingual LBPP",
+          features=_COLUMNS,
+      ),
+      LBPPConfig(name="default", description="Python LBPP", features=_COLUMNS),
+      LBPPConfig(name="python", description="Python LBPP", features=_COLUMNS),
+      LBPPConfig(name="cpp", description="C++ LBPP", features=_COLUMNS),
+      LBPPConfig(name="go", description="Go LBPP", features=_COLUMNS),
+      LBPPConfig(name="java", description="Java LBPP", features=_COLUMNS),
+      LBPPConfig(name="js", description="JavaScript LBPP", features=_COLUMNS),
+      LBPPConfig(
+          name="javascript", description="JavaScript LBPP", features=_COLUMNS
+      ),
+      LBPPConfig(name="rust", description="JavaScript LBPP", features=_COLUMNS),
+  ]
+  DEFAULT_CONFIG_NAME = "python"
+
+  def _info(self):
+    return self.dataset_info_from_configs(
+        features=tfds.features.FeaturesDict({
+            "task_id": tfds.features.Text(),
+            "language": tfds.features.Text(),
+            "title": tfds.features.Text(),
+            "instruction": tfds.features.Text(),
+            "completion": tfds.features.Text(),
+            "test_file": tfds.features.Text(),
+            "test_list": tfds.features.Sequence(tfds.features.Text()),
+            "signature": tfds.features.Text(),
+            "categories": tfds.features.Sequence(tfds.features.Text()),
+            "test_setup": tfds.features.Text(),
+        }),
+        homepage=_HOMEPAGE,
+        supervised_keys=None,
+    )
+
+  def _split_generators(self, dl_manager):
+    # Map alias to actual language
+    data_loading_name = _LANGUAGE_ALIAS_MAP.get(
+        self.builder_config.name, self.builder_config.name
+    )
+    hf_url_prefix = (
+        "https://huggingface.co/datasets/CohereForAI/lbpp/resolve/main/"
+    )
+    if data_loading_name in _ALL_LANGUAGE_ALIASES:
+      # Download all languages
+      download_targets = [
+          f"{hf_url_prefix}{lang}/test.parquet" for lang in _LANGUAGES
+      ]
+    else:
+      download_targets = [f"{hf_url_prefix}{data_loading_name}/test.parquet"]
+
+    downloaded_files = dl_manager.download(download_targets)
+
+    return [
+        tfds.core.SplitGenerator(
+            name=tfds.Split.TEST,
+            gen_kwargs={
+                "filepaths": downloaded_files,
+            },
+        )
+    ]
+
+  def _generate_examples(self, filepaths: list[str]):
+    key = 0
+    for filepath in filepaths:
+      df = pd.read_parquet(filepath)
+      for line in df.to_dict(orient="records"):
+        yield key, {
+            "task_id": line["task_id"],
+            "language": line["language"],
+            "title": line["title"],
+            "instruction": line["instruction"],
+            "completion": decode_str(line["completion"]),
+            "test_file": decode_str(line["test_file"]),
+            "test_list": decode_str(line["test_list"]),
+            "signature": line["signature"] or "",
+            "categories": line["categories"],
+            "test_setup": decode_str(line["test_setup"]),
+        }
+        key += 1

{tfds_nightly-4.9.8.dev202504110044.dist-info → tfds_nightly-4.9.8.dev202504130103.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tfds-nightly
-Version: 4.9.8.dev202504110044
+Version: 4.9.8.dev202504130103
 Summary: tensorflow/datasets is a library of datasets ready to use with TensorFlow.
 Home-page: https://github.com/tensorflow/datasets
 Download-URL: https://github.com/tensorflow/datasets/tags

{tfds_nightly-4.9.8.dev202504110044.dist-info → tfds_nightly-4.9.8.dev202504130103.dist-info}/RECORD

@@ -818,6 +818,10 @@ tensorflow_datasets/datasets/lambada/__init__.py,sha256=UCI6C5ryp7VdOk-mbtcPxp6J
 tensorflow_datasets/datasets/lambada/checksums.tsv,sha256=EgETmQzHlYMzJjZ5u71Pho16UdrbagX34a5jcpgmSmU,172
 tensorflow_datasets/datasets/lambada/lambada_dataset_builder.py,sha256=BNYLj9XI0pdwm1c7wO5ZqJ7BMbMdnCmUr44nX2F8nxE,2269
 tensorflow_datasets/datasets/lambada/lambada_dataset_builder_test.py,sha256=gJwJQ6CJYOt6QlO5c5-VALf804ByrDMwf4dGjOTjE5Y,1070
+tensorflow_datasets/datasets/lbpp/CITATIONS.bib,sha256=-GCp4MjOuqbVCM6scAoXKhNVuRblq-9W1gJQ304FvPU,835
+tensorflow_datasets/datasets/lbpp/README.md,sha256=g1DwOdrH-6todsxwhdnRoJGkB2NzXw5Mq-CgJ-Lsnjo,372
+tensorflow_datasets/datasets/lbpp/__init__.py,sha256=UCI6C5ryp7VdOk-mbtcPxp6JjEup1pI68YhAiayrHjc,612
+tensorflow_datasets/datasets/lbpp/lbpp_dataset_builder.py,sha256=OAG0Pm-jAdb37QB4Al70uXzu7KLqq5mXtT2EirzFWAE,5835
 tensorflow_datasets/datasets/lfw/CITATIONS.bib,sha256=hYwRfGh52dQKV9Z4EX0ec-esXVWaOY3d3g-ulDGYyKQ,340
 tensorflow_datasets/datasets/lfw/README.md,sha256=SKbgr4y6FxNy2Abei2kk2VQ6N1a7MHKcLOhnTGhLI3I,98
 tensorflow_datasets/datasets/lfw/TAGS.txt,sha256=3k_aYCPHQfbOnta3Z41F8BFfpHbzanXx2ZLL7h59TpU,46
@@ -2455,10 +2459,10 @@ tensorflow_datasets/vision_language/wit/wit_test.py,sha256=DJ687VN9hAp6SLXnr_P12
 tensorflow_datasets/vision_language/wit_kaggle/__init__.py,sha256=38tJQ73VHz8vOJn-AyZh2we2YJucbSRIgmgcrsC6bQM,719
 tensorflow_datasets/vision_language/wit_kaggle/wit_kaggle.py,sha256=k-7YD1SGr5bASfdR2_09rrqz-8cpWdIcBWWEXhCvzuk,16903
 tensorflow_datasets/vision_language/wit_kaggle/wit_kaggle_test.py,sha256=vLlluBW77ASNVC4ix7t8idkSUBI6q1-B7zmRV_ICCQM,1778
-tfds_nightly-4.9.8.dev202504110044.dist-info/licenses/AUTHORS,sha256=nvBG4WwfgjuOu1oZkuQKw9kg7X6rve679ObS-YDDmXg,309
-tfds_nightly-4.9.8.dev202504110044.dist-info/licenses/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
-tfds_nightly-4.9.8.dev202504110044.dist-info/METADATA,sha256=H17CedG4r0iROeeuCajllqeuV30nomA-6fC7rTODtjQ,11879
-tfds_nightly-4.9.8.dev202504110044.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-tfds_nightly-4.9.8.dev202504110044.dist-info/entry_points.txt,sha256=eHEL7nF5y1uCY2FgkuYIdE062epJXlAQTSdq89px4p4,73
-tfds_nightly-4.9.8.dev202504110044.dist-info/top_level.txt,sha256=bAevmk9209s_oxVZVlN6hSDIVS423qrMQvmcWSvW4do,20
-tfds_nightly-4.9.8.dev202504110044.dist-info/RECORD,,
+tfds_nightly-4.9.8.dev202504130103.dist-info/licenses/AUTHORS,sha256=nvBG4WwfgjuOu1oZkuQKw9kg7X6rve679ObS-YDDmXg,309
+tfds_nightly-4.9.8.dev202504130103.dist-info/licenses/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
+tfds_nightly-4.9.8.dev202504130103.dist-info/METADATA,sha256=Xb2ci-GJ8aIVhADL-FVI0VIfcvUCVEUiPDRUNdZTGhI,11879
+tfds_nightly-4.9.8.dev202504130103.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+tfds_nightly-4.9.8.dev202504130103.dist-info/entry_points.txt,sha256=eHEL7nF5y1uCY2FgkuYIdE062epJXlAQTSdq89px4p4,73
+tfds_nightly-4.9.8.dev202504130103.dist-info/top_level.txt,sha256=bAevmk9209s_oxVZVlN6hSDIVS423qrMQvmcWSvW4do,20
+tfds_nightly-4.9.8.dev202504130103.dist-info/RECORD,,