INDEX

Explanations

references to educational or informative publications and events, specifically mentioning National Geographic and TED

oai_token-act-pair · gpt-3.5-turbo

references to well-known publications and media outlets

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

qqa

-0.76

rosc

-0.73

 disenfranch

-0.68

oud

-0.66

estic

-0.66

alities

-0.64

oided

-0.64

alam

-0.63

ammy

-0.63

sembly

-0.63

POSITIVE LOGITS

Lens

0.79

 Magazine

0.78

NEWS

0.77

 Channel

0.77

 magazine

0.74

Char

0.72

 Taste

0.72

Tips

0.71

ischer

0.70

 News

0.70

Activations Density 0.029%

references to educational or informative publications and events, specifically mentioning National Geographic and TED

references to well-known publications and media outlets

No Comments

No Known Activations