INDEX

Explanations

terms related to illegal or prohibited actions

oai_token-act-pair · gpt-3.5-turbo

mentions of unlawful activities and physicians

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

oday

-0.90

ocular

-0.90

zsche

-0.81

yrinth

-0.80

tered

-0.78

algia

-0.77

pson

-0.76

allo

-0.76

ago

-0.75

tering

-0.73

POSITIVE LOGITS

theless

0.96

asus

0.76

 actionGroup

0.71

EED

0.69

ONSORED

0.69

sburgh

0.66

ignty

0.64

llor

0.64

FUL

0.61

 motions

0.60

Activations Density 0.079%

terms related to illegal or prohibited actions

mentions of unlawful activities and physicians

No Comments

No Known Activations