Hugging Face's logo

ruotian
/
ConClick-8B

Image-Text-to-Text
PEFT
Safetensors
gui-grounding
computer-use
qwen3-vl
lora
conclick
Model card Files
xet
 Community
1

Add link to paper

#1
by nielsr HF Staff - opened
base: refs/heads/main
←
from: refs/pr/1
Discussion Files changed
+4
-4
Files changed (1) hide show
  1. README.md +4 -4
README.md CHANGED
@@ -1,10 +1,9 @@
1
  ---
2
- license: apache-2.0
3
- library_name: peft
4
  base_model: Qwen/Qwen3-VL-8B-Instruct
5
- base_model_relation: adapter
6
  datasets:
7
  - ruotian/ConClick-Data
 
 
8
  pipeline_tag: image-text-to-text
9
  tags:
10
  - gui-grounding
@@ -12,11 +11,12 @@ tags:
12
  - qwen3-vl
13
  - lora
14
  - conclick
 
15
  ---
16
 
17
  # ConClick-8B
18
 
19
- Model for **ConClick: Element-wise Contrastive Learning Improves GUI Grounding**. ConClick trains on target–distractor examples with coordinate supervision, listwise ranking, and pairwise margin losses.
20
  | Inference | ScreenSpot-Pro | MMBench-GUI L2 | OSWorld-G |
21
  |---|---:|---:|---:|
22
  | Direct | 66.03 | 86.28 | 70.20 |
 
1
  ---
 
 
2
  base_model: Qwen/Qwen3-VL-8B-Instruct
 
3
  datasets:
4
  - ruotian/ConClick-Data
5
+ library_name: peft
6
+ license: apache-2.0
7
  pipeline_tag: image-text-to-text
8
  tags:
9
  - gui-grounding
 
11
  - qwen3-vl
12
  - lora
13
  - conclick
14
+ base_model_relation: adapter
15
  ---
16
 
17
  # ConClick-8B
18
 
19
+ Model for **ConClick: Element-wise Contrastive Learning Improves GUI Grounding** ([paper](https://huggingface.co/papers/2610.13268)). ConClick trains on target–distractor examples with coordinate supervision, listwise ranking, and pairwise margin losses.
20
  | Inference | ScreenSpot-Pro | MMBench-GUI L2 | OSWorld-G |
21
  |---|---:|---:|---:|
22
  | Direct | 66.03 | 86.28 | 70.20 |