lightonai/LightOnOCR-2-1B · Discussions

Resources

View closed (6)

RLVR Strategy Suggestions

#30 opened 11 days ago by

Accuracy decreases a lot in gguf conversion

#28 opened 20 days ago by

How it is compared in bahchmarks with new model Paddle ocr 1.5 ?

#27 opened 23 days ago by

How do you calculate inference speed on OlmOCR-Bench?

#26 opened 24 days ago by

Throughput is slower than PadlleOCR-VL

#25 opened 26 days ago by

Regarding reasoning capability on image

#24 opened 26 days ago by

The exact vocab size of the model

#22 opened about 1 month ago by

abdullahamlwakeb

Future suggestion about using available text layers

#21 opened about 1 month ago by

Prompt of Demo web

#20 opened about 1 month ago by

whats the best settings to run using vllm at very high concurrency

#19 opened about 1 month ago by

Fine-tuning for table bounding box extraction

#18 opened about 1 month ago by

Fine-tune for structured extraction

#17 opened about 1 month ago by

x,y coordinates

#16 opened about 1 month ago by

Superdooperhero

Incredible Model, Thank You!

#15 opened about 1 month ago by

Amazing model

#14 opened about 1 month ago by

Experience on finetuning for specific language

#13 opened about 1 month ago by

Fix typo in model architecture class name

#12 opened about 1 month ago by

Apple users fear not you can use MLX too!

#11 opened about 1 month ago by

LightOnOCR-2-1B with Samaritan Hebrew

#10 opened about 1 month ago by

any way to get a shrunk output to reduce tokens?

#9 opened about 1 month ago by

Muliltipage documents

#8 opened about 1 month ago by

Tables as markdown instead of HTML

#6 opened about 1 month ago by

Remove page numbers from the output

#5 opened about 1 month ago by

hallucinations on empty pages

#4 opened about 1 month ago by

Checkboxes

#2 opened about 1 month ago by

Testing OCR using vllm returns nothing but exclamation marks.

#1 opened about 1 month ago by