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

 narrated

-0.67

Pg

-0.65

ppa

-0.63

mentioned

-0.59

 quotations

-0.59

 Syri

-0.56

 mute

-0.56

LM

-0.56

 fault

-0.56

 Merkel

-0.56

POSITIVE LOGITS

aminer

0.88

omaly

0.83

urdue

0.82

imately

0.79

aez

0.79

ilitating

0.79

ãĥĺãĥ©

0.77

rocal

0.74

Ĥª

0.73

TOR

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.