INDEX

Explanations

terms related to security and terrorism concerns

oai_token-act-pair · gpt-3.5-turbo

references to terrorism and threats

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Bra

-0.85

Â¢

-0.85

Doc

-0.78

 Warranty

-0.75

orest

-0.75

River

-0.74

cum

-0.73

 paddle

-0.72

 Proced

-0.70

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.69

POSITIVE LOGITS

 terrorism

2.09

 terrorist

2.05

 terrorists

2.01

 Terrorism

1.92

Terror

1.90

 jihadist

1.86

 terror

1.81

 Terror

1.81

 ISIS

1.79

 ISIL

1.76

Activations Density 0.743%

terms related to security and terrorism concerns

references to terrorism and threats

No Comments

No Known Activations