INDEX

Explanations

mentions of violent incidents or attacks

oai_token-act-pair · gpt-3.5-turbo

references to various types of attacks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ellen

-0.76

Hub

-0.74

vice

-0.72

 Guinness

-0.67

Lot

-0.66

 Lifetime

-0.66

 Bachelor

-0.65

rolet

-0.65

anse

-0.64

 Cinderella

-0.64

POSITIVE LOGITS

 attacks

3.62

attacks

2.97

 Attacks

2.91

 attack

2.56

 assaults

2.35

 attackers

2.25

attack

2.19

 Attack

1.90

Attack

1.85

 attacked

1.84

Activations Density 0.018%

mentions of violent incidents or attacks

references to various types of attacks

No Comments

No Known Activations