INDEX

Explanations

controversial or notable attributes or features associated with various entities such as people, organizations, or cities

oai_token-act-pair · gpt-3.5-turbo

references to well-known individuals and the attributes associated with them

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Rasm

-0.72

LET

-0.71

 Worse

-0.70

soever

-0.67

CHAPTER

-0.66

river

-0.65

ÑĮ

-0.64

Copyright

-0.63

trl

-0.63

VI

-0.63

POSITIVE LOGITS

own

1.22

 propensity

1.09

 portrayal

1.03

 tendency

1.03

 penchant

1.02

 unorthodox

1.02

 ability

1.00

 willingness

1.00

 inability

0.98

 versatility

0.98

Activations Density 0.139%

controversial or notable attributes or features associated with various entities such as people, organizations, or cities

references to well-known individuals and the attributes associated with them

No Comments

No Known Activations