INDEX

Explanations

the term "perpetrator" or similar words indicating someone responsible for a harmful action

oai_token-act-pair · gpt-3.5-turbo

terms related to criminal actions and the individuals who commit them

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

yip

-0.76

andel

-0.71

mel

-0.69

psey

-0.68

ERAL

-0.66

 Plat

-0.66

zl

-0.66

binding

-0.62

rients

-0.62

 Pyth

-0.61

POSITIVE LOGITS

 perpetrated

0.99

 committed

0.78

 perpetrator

0.77

 perpetrators

0.75

 perpet

0.73

 thereof

0.72

 inflicted

0.72

ãĥĩ

0.71

orst

0.71

 offender

0.70

Activations Density 0.035%

the term "perpetrator" or similar words indicating someone responsible for a harmful action

terms related to criminal actions and the individuals who commit them

No Comments

No Known Activations