INDEX

Explanations

words related to human or humanoid figures or representations

oai_token-act-pair · gpt-3.5-turbo

references to significant individuals or roles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

oise

-0.71

umenthal

-0.70

esis

-0.69

Policy

-0.68

velength

-0.64

izabeth

-0.63

iott

-0.63

itcher

-0.63

 Ples

-0.62

é¾

-0.61

POSITIVE LOGITS

head

1.21

heads

1.20

 skating

1.12

 prominently

1.03

downs

0.83

doms

0.80

books

0.76

enance

0.76

awa

0.76

hig

0.75

Activations Density 0.030%

words related to human or humanoid figures or representations

references to significant individuals or roles

No Comments

No Known Activations