INDEX

Explanations

phrases related to political figures or actions

oai_token-act-pair · gpt-3.5-turbo

nouns related to roles, positions, and professions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Doors

-0.71

earable

-0.68

[|

-0.67

doors

-0.65

angan

-0.64

 everlasting

-0.62

cour

-0.62

fml

-0.62

Content

-0.60

 Amendments

-0.60

POSITIVE LOGITS

glers

0.80

rers

0.77

lier

0.71

arers

0.71

ifier

0.70

ifiers

0.69

kit

0.66

rator

0.66

 advocate

0.66

 cares

0.64

Activations Density 0.647%

phrases related to political figures or actions

nouns related to roles, positions, and professions

No Comments

No Known Activations