INDEX

Explanations

references to discriminatory terms, in this case, the term "jew hater"

oai_token-act-pair · gpt-3.5-turbo

terms related to anti-Semitism

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 blogs

-0.63

blog

-0.61

 wiser

-0.60

bats

-0.60

blogs

-0.59

fid

-0.58

 booth

-0.58

 Berman

-0.57

 withdrawing

-0.57

 booths

-0.57

POSITIVE LOGITS

0.81

anship

0.79

netflix

0.77

yip

0.77

uilt

0.74

_-_

0.73

Britain

0.69

Reviewer

0.68

orthern

0.67

 miscon

0.66

Activations Density 0.000%

references to discriminatory terms, in this case, the term "jew hater"

terms related to anti-Semitism

No Comments

No Known Activations