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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 sshd

-0.73

yson

-0.71

ologically

-0.69

Reviewer

-0.69

 Nicarag

-0.66

ously

-0.64

¬¼

-0.63

 Morning

-0.62

kr

-0.62

Morning

-0.62

POSITIVE LOGITS

FILE

0.73

store

0.69

orted

0.64

allery

0.63

canon

0.62

Haz

0.62

iour

0.60

alph

0.59

ography

0.59

dir

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.