INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

amaz

-0.72

 dilig

-0.69

Magikarp

-0.67

NCT

-0.67

 tast

-0.67

 adolesc

-0.66

 sidel

-0.65

dro

-0.64

 tremend

-0.62

 uncomp

-0.62

POSITIVE LOGITS

het

0.93

ower

0.80

anism

0.77

lyn

0.77

athe

0.75

kee

0.74

pell

0.71

berra

0.69

nee

0.68

adian

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.