INDEX

Explanations

mentions of the politician Dianne Feinstein

oai_token-act-pair · gpt-3.5-turbo

references to a specific individual, Dianne Feinstein

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

IFrame

Link

Negative Logits

urgical

-0.74

ocation

-0.72

 Corpse

-0.71

irgin

-0.69

 Grail

-0.67

akeru

-0.67

icked

-0.65

urgy

-0.65

anked

-0.65

umps

-0.64

POSITIVE LOGITS

 Feinstein

1.15

 Dianne

0.95

bour

0.74

+++

0.70

oslov

0.69

Connor

0.68

UGH

0.68

instein

0.67

++

0.66

 unfocusedRange

0.66

Activations Density 0.021%

mentions of the politician Dianne Feinstein

references to a specific individual, Dianne Feinstein

No Known Activations