Fine-tuned olmOCR 2 beats baseline on transcribing handwritten pottery archives
allen_ai · x · 2026-10-29
Allen AI highlights open-source OCR model olmOCR 2: runs on local hardware, turns scans into searchable text, and can be adapted to specific needs.
For the CENTURIA project, TU Wien and the Austrian Academy of Sciences fine-tuned olmOCR 2 on handwritten pottery records from Carnuntum, Austria. The adapted model made fewer transcription errors on held-out records than the unadapted one, and the team now aims to apply it to the full archive.
Related event: Ai2 Open-Sources olmOCR 2 for Historical Document Transcription(2 posts)→
More from Multimodal
- Reve publishes technical report on training infra that scaled to 10K GPUs with 10x fewer chips — cantrell · 2026-10-29
- Kling 4.0 video model lands on Cloudflare AI Gateway at $0.157/sec for 720p — ritakozlov · 2026-10-29
- AI-generated Star Wars-style game demo shows a flyable desert world with TIE fighters — jasonkneen · 2026-10-29
- LiquidAI's LFM3-Audio-2B audio model trends on Hugging Face — LiquidAI · 2026-10-29
- Generating pop-art Halloween portraits with Flora and a vibe board — round · 2026-10-29
- Generating K-Pop music videos with a Claude Music Video skill, no MCP needed — AIandDesign · 2026-10-29