INDEX

Explanations

references to a person named "Boss"

oai_token-act-pair · gpt-3.5-turbo

mentions of "boss" in various contexts

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

WAYS

-0.81

ACTED

-0.74

CTR

-0.70

âĢ¢âĢ¢âĢ¢âĢ¢

-0.70

OPE

-0.68

VERTIS

-0.67

igenous

-0.66

ances

-0.64

 Coch

-0.64

Pwr

-0.63

POSITIVE LOGITS

anova

0.99

buster

0.93

es

0.91

hatt

0.87

Boss

0.80

erker

0.80

arella

0.79

unin

0.77

enhagen

0.75

zilla

0.74

Activations Density 0.017%

references to a person named "Boss"

mentions of "boss" in various contexts

No Comments

No Known Activations