INDEX

Explanations

phrases related to people and social interactions

oai_token-act-pair · gpt-3.5-turbo

references to various groups of people, often in negative or stereotypical contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Pyr

-0.52

Cur

-0.49

IER

-0.48

Byr

-0.48

Prim

-0.47

 Grind

-0.47

igor

-0.47

 incumbent

-0.47

 Vulcan

-0.47

 Ranger

-0.46

POSITIVE LOGITS

 rejoice

0.85

 unite

0.76

hate

0.75

 beware

0.74

sue

0.73

 adore

0.73

 prefer

0.71

 disapprove

0.70

 dont

0.70

paces

0.69

Activations Density 0.204%

phrases related to people and social interactions

references to various groups of people, often in negative or stereotypical contexts

No Comments

No Known Activations