Update README.md

README.md

@@ -20,15 +20,22 @@ dataset_info:
     sequence: list
 ---
 
-# MOMIJI Dataset Card
+# Dataset Card for MOMIJI
 
 ## Dataset Description
 
 - **GitHub Repository:** (Coming soon)
+- **Paper:** (Coming soon)
 
 **MOMIJI** (**M**odern **O**pen **M**ult**i**modal **J**apanese f**i**ltered Dataset) is a large-scale, carefully curated public dataset of image-text–interleaved web documents. The dataset was extracted from Common Crawl dumps covering February 2024 – January 2025 and contains roughly **56 million** Japanese documents, **110 billion** characters, and **2.49 billion** images. Details of the collection and filtering pipeline will be described in a forthcoming paper.
 
-Image-text–interleaved data is widely used to train large-scale vision-language models (LVLMs) such as [LLaVA-OneVision](https://arxiv.org/abs/2408.03326), [Idefics 2](https://arxiv.org/abs/2405.02246), [NVILA](https://arxiv.org/abs/2412.04468), and [Qwen 2.5-VL](https://arxiv.org/abs/2502.13923). Using MOMIJI, we trained **[Heron-NVILA-Lite](https://huggingface.co/turing-motors/Heron-NVILA-Lite-15B)**.
+Image-text–interleaved data is generally used to train large vision-language models (LVLMs) such as [LLaVA-OneVision](https://arxiv.org/abs/2408.03326), [Idefics 2](https://arxiv.org/abs/2405.02246), [NVILA](https://arxiv.org/abs/2412.04468), and [Qwen 2.5-VL](https://arxiv.org/abs/2502.13923). Using MOMIJI, we trained our proposed model **[Heron-NVILA-Lite](https://huggingface.co/turing-motors/Heron-NVILA-Lite-15B)**.
+
+Following [OBELICS](https://huggingface.co/datasets/HuggingFaceM4/OBELICS), we provide an [interactive visualization](https://atlas.nomic.ai/data/onely7/momiji-1m/map) that allows users to explore the contents of MOMIJI. The map shows a subset of 1M of the 56M documents.  
+
+[![MOMIJI Nomic map](assets/nomic_map_momiji.png)](https://atlas.nomic.ai/data/onely7/momiji-1m/map)
+
+> **Warning and Disclaimer:** This content may unintentionally include expressions or information that some may find inappropriate. Please view it at your own discretion and responsibility.
 
 ## Data Fields
 
@@ -91,11 +98,11 @@ MOMIJI contains about **56 million** Japanese web documents. Because images are
 
 Below is a bar chart of document counts by number of images (documents with ≥ 30 images are omitted for readability):
 
-<img src="https://cdn-uploads.huggingface.co/production/uploads/67297c0bb5f7939284345463/2LZP2cVh-Qc-0PBeWQnKC.png" width="90%">
+<img src="assets/momiji_document_count_by_num_of_images.png" width="90%">
 
 ## Content Warning
 
-Although an NSFW filter was applied, some links or text samples may still be disturbing. The dataset is intended for scientific or safety analyses by trained researchers and is **NOT** meant for casual browsing.
+Although an NSFW filter was applied, some links or text samples may still be disturbing. The dataset is intended for scientific or safety analyses by trained researchers.
 
 ## Disclaimer
 
@@ -107,4 +114,4 @@ MOMIJI does **NOT** distribute image binaries—only links and metadata. We are
 
 ## Acknowledgements
 
-This model is based on results obtained from a project, JPNP20017, subsidized by the New Energy and Industrial Technology Development Organization (NEDO).
+This model is based on results obtained from a project, JPNP20017, subsidized by the New Energy and Industrial Technology Development Organization (NEDO).