INDEX

Explanations

mentions of firearms or weapons, particularly focusing on guns

oai_token-act-pair · gpt-3.5-turbo

references to guns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

grad

-0.72

Fair

-0.69

Attempts

-0.66

lihood

-0.65

spect

-0.64

UTE

-0.64

Work

-0.64

Benef

-0.63

Solution

-0.62

eff

-0.61

POSITIVE LOGITS

linger

1.34

 blazing

1.21

hips

1.15

 guns

1.12

mith

1.09

powder

1.06

poons

0.97

hops

0.96

hooting

0.95

hip

0.93

Activations Density 0.016%

mentions of firearms or weapons, particularly focusing on guns

references to guns

No Comments

No Known Activations