INDEX

Explanations

traits or characteristics with positive or negative connotations such as "all-star", "criminal", "professional", "dark", "competitive", "dumb", "visual", "Swedish residency", "direct", "concrete", "moral", "liberal", "diplomatic", "

oai_token-act-pair · gpt-3.5-turbo

keywords related to criminality and professional qualifications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

hander

-0.82

href

-0.78

rame

-0.72

xtap

-0.71

hooting

-0.70

hemer

-0.69

tyard

-0.69

onement

-0.66

wark

-0.66

hower

-0.65

POSITIVE LOGITS

 aspirations

1.26

 ambitions

1.22

 tendencies

1.21

 intentions

1.19

 privileges

1.14

 backgrounds

1.12

 sensibilities

1.09

 capabilities

1.07

 ties

1.06

 responsibilities

1.05

Activations Density 0.335%

traits or characteristics with positive or negative connotations such as "all-star", "criminal", "professional", "dark", "competitive", "dumb", "visual", "Swedish residency", "direct", "concrete", "moral", "liberal", "diplomatic", "

keywords related to criminality and professional qualifications

No Comments

No Known Activations

traits or characteristics with positive or negative connotations such as "all-star", "criminal", "professional", "dark", "competitive", "dumb", "visual", "Swedish residency", "direct", "concrete", "moral", "liberal", "diplomatic", "

keywords related to criminality and professional qualifications

No Comments

No Known Activations