INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

displayText

-0.83

brance

-0.72

 constructive

-0.67

imeter

-0.67

OUP

-0.67

 procedure

-0.66

ãĥ¼ãĥ«

-0.63

 default

-0.63

ãĥ¤

-0.62

STER

-0.62

POSITIVE LOGITS

absor

0.81

cap

0.74

aples

0.71

0.70

ockey

0.69

mate

0.68

region

0.67

umblr

0.67

eligible

0.66

ulia

0.66

Activations Density 0.000%

No Known Activations

This feature has no known activations.