INDEX

Explanations

references to small, miniature versions of objects or spaces

oai_token-act-pair · gpt-3.5-turbo

mentions of "mini" or miniature concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

âķĲâķĲ

-0.78

Ö¼

-0.77

xual

-0.70

yer

-0.69

lain

-0.67

rer

-0.66

 defe

-0.63

itbart

-0.63

mingham

-0.63

 depl

-0.61

POSITIVE LOGITS

ature

1.32

atur

1.21

atures

1.07

ocular

0.88

eco

0.83

 sized

0.79

oby

0.77

ota

0.77

atomic

0.76

isode

0.75

Activations Density 0.010%

references to small, miniature versions of objects or spaces

mentions of "mini" or miniature concepts

No Comments

No Known Activations