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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

vale

-0.75

always

-0.72

toe

-0.69

upon

-0.68

uther

-0.66

aples

-0.65

agall

-0.65

 Keynes

-0.65

matter

-0.64

closed

-0.64

POSITIVE LOGITS

 corrid

0.72

ivity

0.69

 Intro

0.68

 speech

0.65

 localization

0.65

 selfie

0.64

 branding

0.64

Þ

0.62

 imaging

0.62

htaking

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.