Sumtablets Merged: Cuneiform Tablet Text and Images for Vision-Language Models

A multimodal training corpus combining cuneiform tablet text records with tablet photography and lineart. The dataset was merged, deduplicated, and quality-filtered by author TRACCERR and was last updated on June 2, 2026. It is structured for fine-tuning vision-language models, specifically targeting the Qwen3-VL-8B-Instruct model.

Use Cases

• Fine-tuning vision-language models based on aligned text and image data of cuneiform tablets.
• Training models for multimodal classification of ancient artifacts based on tablet photography and lineart.
• Developing tools for the automated transcription or translation of cuneiform text based on visual input.
• Conducting research on the intersection of computer vision and historical linguistics using structured multimodal records.

Strengths

• Dataset has undergone merging, deduplication, and quality-filtering processes.
• Data is explicitly structured for a specific model fine-tuning task (Qwen3-VL-8B-Instruct).
• License information is provided (CC-BY-4.0, with the most restrictive source license applying).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and dataset size are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface, author TRACCERR

Collection Method

Merged, deduplicated, and quality-filtered from source datasets containing cuneiform records.

Freshness

Last updated 2026-06-02 17:59:24; freshness should be verified.