INDEX

Explanations

words related to aggression or conflict

oai_token-act-pair · gpt-3.5-turbo

references to attention or focus

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

 shorth

-0.65

 Giuliani

-0.64

 Cuomo

-0.60

ONE

-0.60

ãĤ¬

-0.60

ONES

-0.57

OPLE

-0.57

 Galile

-0.57

 Bosnia

-0.57

 Tunisia

-0.56

POSITIVE LOGITS

ractive

1.50

itudinal

1.48

itudes

1.47

ention

1.46

orneys

1.45

achment

1.41

itude

1.39

raction

1.35

acking

1.27

ached

1.26

Activations Density 0.019%

words related to aggression or conflict

references to attention or focus

No Comments

No Known Activations