first try oposite streaming, then deal with unverfied code, as the un…

…verfied swallows the loading problem Signed-off-by: dafnapension <[email protected]>
IBM · Feb 8, 2025 · e37b7e9 · e37b7e9
1 parent 6f1b886
commit e37b7e9
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 17 deletions.
diff --git a/performance/compare_benchmark_performance_results.py b/performance/compare_benchmark_performance_results.py
@@ -70,8 +70,8 @@
         "\n**Warning**: Performance degradation in Dataset Generation and/or Evaluation exceeds 5%!"
     )
     print(
-        "Explore branch performance via 'python performance/bluebench_profiler.py --output_file=<path to json file>',"
-        "followed by 'snakeviz <the performance.prof file specified in the output json file>'."
+        "Explore branch performance via 'python performance/bluebench_profiler.py --output_file=``path to json file``',"
+        "followed by 'snakeviz ``the performance.prof file specified in the output json file``'."
     )
     sys.exit(1)
 

diff --git a/src/unitxt/loaders.py b/src/unitxt/loaders.py
@@ -312,20 +312,28 @@ def load_dataset(
                         for k in dataset.keys():
                             next(iter(dataset[k]))
                             break
-                except ValueError as e:
-                    if "trust_remote_code" in str(e):
-                        raise ValueError(
-                            f"{self.__class__.__name__} cannot run remote code from huggingface without setting unitxt.settings.allow_unverified_code=True or by setting environment variable: UNITXT_ALLOW_UNVERIFIED_CODE."
-                        ) from e
 
                 except:
-                    current_streaming = kwargs["streaming"]
-                    logger.info(
-                        f"needed to swap streaming from {current_streaming} to {not current_streaming} for path {self.path}"
-                    )
-                    # try the opposite way of streaming
-                    kwargs["streaming"] = not kwargs["streaming"]
-                    dataset = hf_load_dataset(**kwargs)
+                    try:
+                        current_streaming = kwargs["streaming"]
+                        logger.info(
+                            f"needed to swap streaming from {current_streaming} to {not current_streaming} for path {self.path}"
+                        )
+                        # try the opposite way of streaming
+                        kwargs["streaming"] = not kwargs["streaming"]
+                        dataset = hf_load_dataset(**kwargs)
+                        if isinstance(dataset, (Dataset, IterableDataset)):
+                            next(iter(dataset))
+                        else:
+                            for k in dataset.keys():
+                                next(iter(dataset[k]))
+                                break
+
+                    except ValueError as e:
+                        if "trust_remote_code" in str(e):
+                            raise ValueError(
+                                f"{self.__class__.__name__} cannot run remote code from huggingface without setting unitxt.settings.allow_unverified_code=True or by setting environment variable: UNITXT_ALLOW_UNVERIFIED_CODE."
+                            ) from e
 
             if self.filtering_lambda is not None:
                 dataset = dataset.filter(eval(self.filtering_lambda))

diff --git a/utils/.secrets.baseline b/utils/.secrets.baseline
@@ -151,7 +151,7 @@
         "filename": "src/unitxt/loaders.py",
         "hashed_secret": "840268f77a57d5553add023cfa8a4d1535f49742",
         "is_verified": false,
-        "line_number": 621,
+        "line_number": 629,
         "is_secret": false
       }
     ],
@@ -184,5 +184,5 @@
       }
     ]
   },
-  "generated_at": "2025-02-08T13:05:57Z"
+  "generated_at": "2025-02-08T13:56:45Z"
 }
diff --git a/utils/prepare_all_artifacts.py b/utils/prepare_all_artifacts.py
@@ -42,7 +42,7 @@ def main():
     os.environ["UNITXT_TEST_CARD_DISABLE"] = "True"
     os.environ["UNITXT_TEST_METRIC_DISABLE"] = "True"
     os.environ["UNITXT_ALLOW_UNVERIFIED_CODE"] = "True"
-    os.environ["UNITXT_SKIP_ARTIFACTS_PREPARE_AND_VERIFY"] = "True"
+    os.environ["UNITXT_SKIP_ARTIFACTS_PREPARE_AND_VERIFY"] = "False"
     logger.info("*" * 100)
     logger.info("*" * 100)
     logger.info(