INDEX

Explanations

names of people, potentially from news articles

oai_token-act-pair · gpt-3.5-turbo

names of individuals, particularly prominent figures and officials

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

PDATE

-0.83

GoldMagikarp

-0.76

 rall

-0.75

 looph

-0.74

translation

-0.69

 tyre

-0.68

uesday

-0.67

ccording

-0.64

undai

-0.63

Tact

-0.62

POSITIVE LOGITS

 herself

1.47

ova

1.05

eva

1.02

 husband

0.93

kaya

0.86

 Actress

0.85

 breasts

0.84

 hijab

0.83

 Christina

0.82

 boyfriend

0.81

Activations Density 0.283%

names of people, potentially from news articles

names of individuals, particularly prominent figures and officials

No Comments

No Known Activations