INDEX

Explanations

terms related to violence or aggression

oai_token-act-pair · gpt-3.5-turbo

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

leaf

-0.82

FU

-0.72

ploma

-0.71

BU

-0.70

cript

-0.70

 Unlimited

-0.69

hner

-0.69

ource

-0.69

OPLE

-0.68

Script

-0.68

POSITIVE LOGITS

ized

0.97

 assault

0.92

 assaults

0.87

ified

0.86

 retribution

0.85

 killers

0.84

 beasts

0.83

 punishments

0.83

 murdering

0.82

izing

0.82

Activations Density 0.021%

terms related to violence or aggression

No Comments

No Known Activations