Chandra 2.7 detail thread: rotated pages 87.3%, forms and non-English gains
VikParuchuri · x · 2026-11-17
VikParuchuri details remaining Chandra 2.7 gains: rotated pages balanced pass rate 81.4% → 87.3%, forms accurate 92.3% → 93.7%, non-English pages 86.7% → 87.9%.
Related event: Datalab releases open-source OCR model Chandra 2.7 with big table gains(4 posts)→
More from Multimodal
- Building a game-ready castle pack entirely with AI: the pitfalls of AI 3D modeling — Oaker_at · 2026-11-17
- After DALL-E Shutdown Forced a Pipeline Rewrite: Four Questions for Picking Generative Models — Al_Grigor · 2026-11-17
- 80s-style fantasy AI video dubbed better than movies — mark_k · 2026-11-17
- New audio encoders called a huge leap: embeddings far more informative, audio-CoCa teased — wightmanr · 2026-11-17
- The Burrito Test: one prompt stress-tests 25 image generation models side by side — generativist · 2026-11-17
- Nano Banana keeps failing face identity: a stylized 3D face-swap pipeline question — LocksmithOk4723 · 2026-11-17