INDEX

Explanations

different types or categories of items

oai_token-act-pair · gpt-3.5-turbo

references to different categories or classifications

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

ITNESS

-0.65

ometown

-0.65

 Morning

-0.64

NING

-0.61

 Guant

-0.59

yer

-0.58

ITED

-0.58

IRO

-0.56

onel

-0.56

Rio

-0.56

POSITIVE LOGITS

etting

1.37

etter

1.35

faces

1.02

otypes

0.94

classes

0.92

otyp

0.89

ahead

0.86

face

0.86

isms

0.85

otype

0.82

Activations Density 0.008%

different types or categories of items

references to different categories or classifications

No Comments

No Known Activations