INDEX

Explanations

the word "downs" with varying levels of intensity

oai_token-act-pair · gpt-3.5-turbo

references to the concept of drawbacks or negative aspects

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Gears

-0.70

kun

-0.70

Asia

-0.67

 Portuguese

-0.66

eur

-0.66

 Geneva

-0.65

 Lisbon

-0.64

 Liberia

-0.64

Finish

-0.64

Footnote

-0.64

POSITIVE LOGITS

 downs

1.39

sembly

1.03

eatures

0.98

 Downs

0.90

downs

0.89

stairs

0.86

ayer

0.86

osate

0.83

quished

0.83

 redes

0.80

Activations Density 0.006%

the word "downs" with varying levels of intensity

references to the concept of drawbacks or negative aspects

No Comments

No Known Activations