Hugging Face's logo

numind
/
NuExtract3

Image-to-Text
Transformers
Safetensors
qwen3_5
image-text-to-text
vision-language
vlm
document-understanding
structured-extraction
information-extraction
ocr
document-to-markdown
markdown
rag
reasoning
multilingual
conversational
Eval Results
Model card Files
xet
 Community
4

Add ExtractBench evaluation results

#4
by boyang-runllama - opened
base: refs/heads/main
←
from: refs/pr/4
Discussion Files changed
+40
-0
Files changed (1) hide show
  1. .eval_results/extractbench.yaml +40 -0
.eval_results/extractbench.yaml ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ - dataset:
2
+ id: llamaindex/ExtractBench
3
+ task_id: mean
4
+ value: 82.35
5
+ date: '2026-08-20'
6
+ source:
7
+ url: https://huggingface.co/datasets/llamaindex/ExtractBench
8
+ name: ExtractBench
9
+ user: boyang-runllama
10
+ notes: "Pipeline name: nuextract3_extract; served checkpoint numind/NuExtract3 on vLLM, one-shot template-guided extraction (the JSON Schema is converted to a NuExtract template)"
11
+ - dataset:
12
+ id: llamaindex/ExtractBench
13
+ task_id: short
14
+ value: 88.06
15
+ date: '2026-08-20'
16
+ source:
17
+ url: https://huggingface.co/datasets/llamaindex/ExtractBench
18
+ name: ExtractBench
19
+ user: boyang-runllama
20
+ notes: "Pipeline name: nuextract3_extract; served checkpoint numind/NuExtract3 on vLLM, one-shot template-guided extraction (the JSON Schema is converted to a NuExtract template)"
21
+ - dataset:
22
+ id: llamaindex/ExtractBench
23
+ task_id: medium
24
+ value: 76.76
25
+ date: '2026-08-20'
26
+ source:
27
+ url: https://huggingface.co/datasets/llamaindex/ExtractBench
28
+ name: ExtractBench
29
+ user: boyang-runllama
30
+ notes: "Pipeline name: nuextract3_extract; served checkpoint numind/NuExtract3 on vLLM, one-shot template-guided extraction (the JSON Schema is converted to a NuExtract template)"
31
+ - dataset:
32
+ id: llamaindex/ExtractBench
33
+ task_id: long
34
+ value: 37.72
35
+ date: '2026-08-20'
36
+ source:
37
+ url: https://huggingface.co/datasets/llamaindex/ExtractBench
38
+ name: ExtractBench
39
+ user: boyang-runllama
40
+ notes: "Pipeline name: nuextract3_extract; served checkpoint numind/NuExtract3 on vLLM, one-shot template-guided extraction (the JSON Schema is converted to a NuExtract template)"