INDEX

Explanations

words related to relationships or associations

oai_token-act-pair · gpt-3.5-turbo

references to romantic or close relationships

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

cale

-0.77

ihil

-0.74

aston

-0.71

 oxide

-0.70

kefeller

-0.70

ibaba

-0.69

anism

-0.69

ombie

-0.69

acco

-0.66

ux

-0.66

POSITIVE LOGITS

 partner

1.01

 partners

0.92

 Partner

0.77

rette

0.76

less

0.68

itute

0.67

 Karin

0.67

ries

0.67

hood

0.65

Cod

0.63

Activations Density 0.018%

words related to relationships or associations

references to romantic or close relationships

No Comments

No Known Activations