naver-clova-ix/donut-base-finetuned-docvqa
Primitive: /extract · Extract ·
Encoder-Decoder
Donut model fine-tuned on DocVQA. It was introduced in the paper OCR-free Document Understanding Transformer by Geewok et al. and first released in this repository.
MultimodalText regions
Overview
Hardware: — drives latency, throughput & cost
| Size | 110M params |
|---|---|
| Tasks | /extract |
| License | mit |
| Concurrency | 16 |
| Latency | 12.0 s |
| Throughput | 65 tok/s |
| Cost | $3.41 /1M tok |
Cost is approximate — computed from list GPU prices; your actual price depends on the provider you deploy SIE with.
Extraction
| Output kinds | text_regions |
|---|---|
| Inputs | image |
| Max sequence length | — |
Benchmarks
DocVQA
Visual question answering on document images
Corpus: 5,188 Queries: 5,188
Quality
anls 0.6350
Performance L4-SPOT b1 c4
Extract 109 tok/s
Extract p50 1.8s
Performance L4 b1 c16
Extract 65 tok/s
Extract p50 12.0s
Compare (0)Compare →