Update MMSearch-Plus dataset with encrypted data files

README.md

@@ -71,7 +71,7 @@ Official repository for the paper "[MMSearch-Plus: Benchmarking Provenance-Aware
 
 **⚠️ Important: This dataset is encrypted to prevent data contamination. However, decryption is handled transparently by the dataset loader.**
 
-### Simple Usage (Recommended)
+### Dataset Usage
 
 Load the dataset with automatic decryption using your canary string:
 
@@ -80,7 +80,7 @@ import os
 from datasets import load_dataset
 
 # Set the canary string (hint: it's the name of this repo)
-os.environ['MMSEARCH_CANARY'] = 'your_canary_string'
+os.environ['MMSEARCH_PLUS'] = 'your_canary_string'
 
 # Load dataset with transparent decryption
 dataset = load_dataset("Cie1/MMSearch-Plus", trust_remote_code=True)
@@ -129,14 +129,14 @@ Evaluation of closed- and open-source MLLMs shows:
 - Models struggle with multi-step visual reasoning and cross-modal information integration 
 
 <p align="center">
-    <img src="https://raw.githubusercontent.com/mmsearch-plus/mmsearch-plus.github.io/main/static/images/teaser.png" width="75%"> <br>
+    <img src="https://raw.githubusercontent.com/mmsearch-plus/mmsearch-plus.github.io/main/static/images/teaser.png" width="80%"> <br>
     The overview of three paradigms for multimodal browsing tasks that demand fine-grained visual reasoning.
 </p>
 
 
 
 <p align="center">
-    <img src="https://raw.githubusercontent.com/mmsearch-plus/mmsearch-plus.github.io/main/static/images/real-teaser.jpg" width="60%"> <br>
+    <img src="https://raw.githubusercontent.com/mmsearch-plus/mmsearch-plus.github.io/main/static/images/real-teaser.jpg" width="80%"> <br>
     The overview of an example trajectory for a task in <b>MMSearch-Plus</b>.
 </p>
 

data-00000-of-00003.arrow

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d303f1ca1a9cd9470d401a538f55e1d4d70f9ca07aed8b2ab8d23f635e59831a
-size 419738728
+oid sha256:5d37bfccc2c49668422470eddaccb1aa3abe8d022b6c36f8f5d3b5eb4a23c9b9
+size 419741304

data-00001-of-00003.arrow

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f070af1555340c0202102f292ef2f7920fccce0434b65c6fd384de407db9b9da
-size 466499832
+oid sha256:731843bf222a4ab7e8facf9a468b5d84fb7b8af98d37dfb9b09bc877d278542a
+size 466502312

data-00002-of-00003.arrow

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7d59c12bdf310ad6399d4f94fc9cda5cabcd3a059e528accb8653a253e466adc
-size 345386360
+oid sha256:25e89cb48e883d9884e9dd2acc0498c89556987385fcb2c0badb8933a15a9f6b
+size 345388808

dataset_info.json

@@ -2,10 +2,6 @@
   "citation": "",
   "description": "",
   "features": {
-    "question": {
-      "dtype": "string",
-      "_type": "Value"
-    },
     "answer": {
       "feature": {
         "dtype": "string",
@@ -13,15 +9,19 @@
       },
       "_type": "Sequence"
     },
-    "num_images": {
-      "dtype": "int64",
+    "encrypted_img_1": {
+      "dtype": "string",
+      "_type": "Value"
+    },
+    "encrypted_img_4": {
+      "dtype": "string",
       "_type": "Value"
     },
     "arxiv_id": {
       "dtype": "string",
       "_type": "Value"
     },
-    "video_url": {
+    "question": {
       "dtype": "string",
       "_type": "Value"
     },
@@ -29,33 +29,49 @@
       "dtype": "string",
       "_type": "Value"
     },
-    "difficulty": {
+    "img_5": {
       "dtype": "string",
       "_type": "Value"
     },
-    "subtask": {
+    "encrypted_img_5": {
       "dtype": "string",
       "_type": "Value"
     },
-    "img_1": {
+    "difficulty": {
       "dtype": "string",
       "_type": "Value"
     },
-    "img_5": {
+    "img_4": {
       "dtype": "string",
       "_type": "Value"
     },
-    "img_4": {
+    "encrypted_img_2": {
       "dtype": "string",
       "_type": "Value"
     },
-    "img_2": {
+    "subtask": {
       "dtype": "string",
       "_type": "Value"
     },
     "img_3": {
       "dtype": "string",
       "_type": "Value"
+    },
+    "video_url": {
+      "dtype": "string",
+      "_type": "Value"
+    },
+    "num_images": {
+      "dtype": "int64",
+      "_type": "Value"
+    },
+    "encrypted_img_3": {
+      "dtype": "string",
+      "_type": "Value"
+    },
+    "img_2": {
+      "dtype": "string",
+      "_type": "Value"
     }
   },
   "homepage": "",

mmsearch_plus.py

@@ -104,7 +104,7 @@ class MmsearchPlus(datasets.GeneratorBasedBuilder):
 
     def _split_generators(self, dl_manager):
         # Get canary from environment variable or kwargs
-        canary = os.environ.get("MMSEARCH_CANARY")
+        canary = os.environ.get("MMSEARCH_PLUS")
 
         # Check if passed in the builder's initialization
         if hasattr(self, 'canary'):
@@ -112,10 +112,10 @@ class MmsearchPlus(datasets.GeneratorBasedBuilder):
 
         if not canary:
             raise ValueError(
-                "Canary string is required for decryption. Either set the MMSEARCH_CANARY "
+                "Canary string is required for decryption. Either set the MMSEARCH_PLUS "
                 "environment variable or pass it via the dataset loading kwargs. "
                 "Example: load_dataset('path/to/dataset', trust_remote_code=True) after setting "
-                "os.environ['MMSEARCH_CANARY'] = 'your_canary_string'"
+                "os.environ['MMSEARCH_PLUS'] = 'your_canary_string'"
             )
 
         # Download files
@@ -156,11 +156,28 @@ class MmsearchPlus(datasets.GeneratorBasedBuilder):
                             decrypted_answers.append(answer)
                     example["answer"] = decrypted_answers
 
-                # Decrypt image fields
-                image_fields = ["img_1", "img_2", "img_3", "img_4", "img_5"]
-                for field in image_fields:
-                    if example.get(field):
-                        example[field] = decrypt_image(example[field], canary)
+                # Decrypt image fields - handle both original and renamed encrypted fields
+                image_field_mapping = {
+                    "encrypted_img_1": "img_1",
+                    "encrypted_img_2": "img_2",
+                    "encrypted_img_3": "img_3",
+                    "encrypted_img_4": "img_4",
+                    "encrypted_img_5": "img_5",
+                    # Also handle original field names for backward compatibility
+                    "img_1": "img_1",
+                    "img_2": "img_2",
+                    "img_3": "img_3",
+                    "img_4": "img_4",
+                    "img_5": "img_5"
+                }
+
+                for encrypted_field, decrypted_field in image_field_mapping.items():
+                    if example.get(encrypted_field):
+                        decrypted_image = decrypt_image(example[encrypted_field], canary)
+                        example[decrypted_field] = decrypted_image
+                        # Remove the encrypted field if it's not the same as decrypted field
+                        if encrypted_field != decrypted_field and encrypted_field in example:
+                            del example[encrypted_field]
 
                 yield key, example
                 key += 1

state.json

@@ -10,7 +10,7 @@
       "filename": "data-00002-of-00003.arrow"
     }
   ],
-  "_fingerprint": "f322dd0e15bea130",
+  "_fingerprint": "c8f35004d8a5d590",
   "_format_columns": null,
   "_format_kwargs": {},
   "_format_type": null,