toxicity-benchmark Workspace – Weights & Biases

Skip to main content

Toxicity

openai_moderation_ds:v0

kaggle_toxic_ds:v0

Model

true_fraction

1

f1

2

recall

3

true_fraction

4

f1

5

recall

6

OpenaiModeration:v7

80.65%

74.31%

90.04%

86.90%

86.99%

83.27%

LlamaGuardModel:v3

78.10%

68.22%

75.67%

57.30%

47.74%

37.07%

CeladonModel:v20

77.38%

60.99%

56.90%

N/A

N/A

N/A

CeladonModel:v21

76.61%

58.32%

52.68%

N/A

N/A

N/A

CeladonModel:v19

76.55%

63.25%

64.94%

N/A

N/A

N/A

CeladonModel:v22

76.55%

57.63%

51.34%

N/A

N/A

N/A

CeladonModel:v24

76.49%

57.48%

51.15%

N/A

N/A

N/A

75.83%

61.41%

61.88%

84.40%

84.55%

81.18%

ToxicSmolLM:v16

74.40%

61.47%

65.71%

66.50%

57.22%

42.59%

CeladonModel:v14

74.05%

65.94%

80.84%

N/A

N/A

N/A

CeladonModel:v17

74.05%

65.72%

80.08%

70.90%

67.85%

58.37%

CeladonModel:v23

74.05%

65.72%

80.08%

N/A

N/A

N/A

CeladonModel:v18

74.05%

65.72%

80.08%

N/A

N/A

N/A

CeladonModel:v16

74.05%

65.72%

80.08%

N/A

N/A

N/A

CeladonModel:v15

74.05%

65.72%

80.08%

N/A

N/A

N/A

ToxicSmolLM:v14

73.99%

55.81%

52.87%

N/A

N/A

N/A

Total Rows: 32