Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Triple Validity Evaluation on Knowledge Graph (KG) Triple Validity (full)
Loading...
68.7
Validity Score (Qwen3-32B)
GraphMERT
30.324
40.287
50.25
60.213
Oct 10, 2025
Validity Score (Qwen3-32B)
Maybe Rate (Qwen3-32B)
No Rate (Qwen3-32B)
Missing Rate (Qwen3-32B)
Validity Score (Gemini 2.0 Flash)
Maybe Rate (Gemini 2.0 Flash)
No Rate (Gemini 2.0 Flash)
Missing Rate (Gemini 2.0 Flash)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Validity Score (Qwen3-32B)
Maybe Rate (Qwen3-32B)
No Rate (Qwen3-32B)
Missing Rate (Qwen3-32B)
Validity Score (Gemini 2.0 Flash)
Maybe Rate (Gemini 2.0 Flash)
No Rate (Gemini 2.0 Flash)
Missing Rate (Gemini 2.0 Flash)
GraphMERT
Model size=80M, #tripl...
2025.10
68.7
18.3
10.8
2.2
81.3
18.5
0.2
10
Grok 4 Fast
Model size=1.7T, #trip...
2025.10
56.9
21.9
21.2
0.1
68.8
30.9
0.3
10
UMLS Seed KG
Model size=N/A, #tripl...
2025.10
53.4
10
34.7
1.9
82.5
16.5
1.1
10
Qwen3-32B (baseline)
Model size=32B, #tripl...
2025.10
43
24.1
31.4
1.5
66
33.6
0.3
10
Qwen3-14B
Model size=14B, #tripl...
2025.10
31.8
21.9
46.2
0.1
53.4
45.7
0.9
10
Feedback
Search any
task
Search any
task