Update README.md

README.md

@@ -1,63 +1,117 @@
 ---
 license: mit
-pretty_name: MMB Counterfactual Dataset
+pretty_name: MMIB Evaluation Dataset
 task_categories:
-  - visual-question-answering
-  - multiple-choice
+- visual-question-answering
+- multiple-choice
 language:
-  - en
+- en
 tags:
-  - vision
-  - language
-  - multimodal
-  - counterfactual
-  - question-answering
-  - synthetic
+- vision
+- language
+- multimodal
+- counterfactual
+- mechanistic-interpretability
+- synthetic
 size_categories:
-  - 1K<n<10K
+- n<1K
+dataset_info:
+  features:
+  - name: original_image
+    dtype: image
+  - name: counterfactual1_image
+    dtype: image
+  - name: counterfactual2_image
+    dtype: image
+  - name: counterfactual1_type
+    dtype: string
+  - name: counterfactual2_type
+    dtype: string
+  - name: counterfactual1_description
+    dtype: string
+  - name: counterfactual2_description
+    dtype: string
+  - name: original_question
+    dtype: string
+  - name: counterfactual1_question
+    dtype: string
+  - name: counterfactual2_question
+    dtype: string
+  - name: original_question_difficulty
+    dtype: string
+  - name: counterfactual1_question_difficulty
+    dtype: string
+  - name: counterfactual2_question_difficulty
+    dtype: string
+  - name: original_image_answer_to_original_question
+    dtype: string
+  - name: original_image_answer_to_cf1_question
+    dtype: string
+  - name: original_image_answer_to_cf2_question
+    dtype: string
+  - name: cf1_image_answer_to_original_question
+    dtype: string
+  - name: cf1_image_answer_to_cf1_question
+    dtype: string
+  - name: cf1_image_answer_to_cf2_question
+    dtype: string
+  - name: cf2_image_answer_to_original_question
+    dtype: string
+  - name: cf2_image_answer_to_cf1_question
+    dtype: string
+  - name: cf2_image_answer_to_cf2_question
+    dtype: string
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data/train-*
 ---
 
-# MMB Counterfactual Dataset
+# Multimodal Mechanistic Interpretability Benchmark (MMIB) Dataset
 
-A counterfactual VQA dataset constructed using the CLEVR blender assets to procedurally generate both negative and normal counter factual VQA images and questions for the Multimodal Benchmark paper.
-## Dataset Structure
 
-This repository contains counterfactual visual question answering data with:
 
-- **Original images** and **counterfactual variants** (modifications to test reasoning)
-- **Questions** for each image variant
-- **Answer matrices** showing how each image answers each question (9 values per scene: 3 images × 3 questions)
+The **MMIB Dataset** is a highly controlled, synthetic vision-language dataset designed to rigorously evaluate mechanistic interpretability (MI) methods in Large Multimodal Models (VLMs). Built upon procedurally generated CLEVR-style assets, this dataset provides exact ground-truth causal pathways to test whether MI techniques (like causal tracing or interchange interventions) localize genuine cognitive circuits or merely identify descriptive correlations.
 
+Unlike standard VQA benchmarks, MMIB uses strict **automated rejection sampling** to eliminate geometric ambiguity, ensuring every spatial and causal relationship is mathematically verifiable. 
+
+## Dataset Structure & Interventions
+
+This dataset is built on a structured intervention triplet for every base scene. Each row provides a complete $3 \times 3$ cross-modal evaluation matrix (3 images $\times$ 3 text queries), allowing researchers to systematically trace cross-modal information flow.
+
+### 1. Semantic Counterfactuals (Causal Reasoning)
+To evaluate the model's internal causal logic, we generate minimal counterfactual pairs where the intervention mathematically guarantees a change in the ground-truth answer ($y' \neq y$).
+* **Image-Based CFs:** 10 targeted 3D scene graph edits (e.g., `change_color`, `change_position`, `relational_flip`) that alter the visual logic while keeping the question fixed.
+* **Text-Based CFs:** Minimal deterministic mutations to the textual query (e.g., swapping "red" for "blue" or "left" for "right") that guarantee an answer flip on the fixed base image. 
+
+### 2. Negative Counterfactuals (Diagnostic Stress Tests)
+To control for basic visual fragility, we generate **Negative Counterfactuals** featuring 8 types of perceptual corruptions (e.g., `add_noise`, `change_lighting`, `apply_fisheye`). These interventions drastically alter the image distribution *without* changing the underlying 3D geometry or ground-truth answer ($y' = y$). They serve as an experimental baseline: if a model fails on these stress tests, its failure on semantic tasks indicates vulnerability to domain shifts rather than flawed causal logic.
+
+## Using the Dataset
 
 ### Loading from Python
 
-After pushing this repository to the Hub, load it with:
+The dataset is hosted in standard Parquet format. You can load it directly into your mechanistic evaluation pipeline using the Hugging Face `datasets` library:
 
 ```python
 from datasets import load_dataset
 
+# Load the MMIB dataset
 ds = load_dataset("scholo/MMB_dataset", split="train")
-print(ds[0])
-```
-
-No `trust_remote_code=True` needed since we use standard Parquet format!
-
-## Directory Structure
-
-```
-MMB-Dataset/
-├── README.md                   # This file
-├── .gitattributes              # Git LFS configuration for images
-├── data/                       # Dataset files (Parquet format)
-│   └── train.parquet           # Main dataset file
-├── Dataset/   # Current dataset run
-│   ├── images/                 # All PNG images (referenced by Parquet)
-│   ├── scenes/                 # JSON scene descriptions (reference)
-│   ├── image_mapping_with_questions.csv  # Original CSV (source)
-│   ├── checkpoint.json         # Run metadata
-│   └── run_metadata.json       # Run metadata
-```
-
-## License
-
-MIT
+
+# Inspect the 3x3 evaluation matrix for the first scene
+print("Base Question:", ds[0]['original_question'])
+print("Base Image -> Base Question Answer:", ds[0]['original_image_answer_to_original_question'])
+print("Semantic CF Image -> Base Question Answer:", ds[0]['cf1_image_answer_to_original_question'])
+(No trust_remote_code=True is required.)Directory StructureMMB-Dataset/
+├── README.md                           # This dataset card
+├── .gitattributes                      # Git LFS configuration
+├── data/                               # Dataset files (Parquet format)
+│   └── train.parquet                   # Main benchmark matrix
+├── Dataset/                            # Raw generation artifacts
+│   ├── images/                         # Uncompressed PNG renders (720x720)
+│   ├── scenes/                         # JSON 3D scene graphs and metadata
+│   ├── image_mapping_with_questions.csv # Source mapping for the 3x3 grid
+│   └── run_metadata.json               # Procedural generation engine parameters
+Application & ProtocolFollowing the rigorous evaluation protocol established in the MMIB paper, interpretability metrics (such as Circuit Performance Ratio, Circuit-Model Distance, and Interchange Intervention Accuracy) should only be computed on samples where the target VLM correctly answers the base question ($a_b = y$). This behavioral filter ensures that the model possesses the causal circuit prior to mechanistic evaluation.LicenseMIT