INDEX

Explanations

words related to violence or intense negative experiences

oai_token-act-pair · gpt-3.5-turbo

references to violent or distressing imagery

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

PLIED

-0.88

Reviewer

-0.88

anol

-0.87

Demand

-0.86

Recommend

-0.78

Rate

-0.77

rador

-0.76

BOOK

-0.75

later

-0.75

CHAT

-0.75

POSITIVE LOGITS

 bloody

0.93

 noses

0.83

 wounds

0.77

 bast

0.77

 slaughter

0.74

 swath

0.74

 prick

0.72

stained

0.72

 swat

0.72

 blood

0.71

Activations Density 0.011%

words related to violence or intense negative experiences

references to violent or distressing imagery

No Comments

No Known Activations