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

¿½

-0.81

ĸļ

-0.72

oice

-0.71

atchewan

-0.70

²¾

-0.70

hester

-0.70

Graphics

-0.68

©¶æ¥µ

-0.68

neau

-0.65

croft

-0.65

POSITIVE LOGITS

 Effect

0.71

historic

0.67

 DRAG

0.65

oft

0.64

 Derby

0.60

Zan

0.60

Dad

0.59

 SIGN

0.59

ulla

0.59

 capit

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.