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

 liking

-0.75

 Fallon

-0.70

 vengeance

-0.65

 retali

-0.64

 Brach

-0.60

Redditor

-0.60

 recomm

-0.60

 behav

-0.59

Geh

-0.58

 morale

-0.58

POSITIVE LOGITS

ierra

0.86

icals

0.81

uti

0.79

jriwal

0.78

celona

0.78

quer

0.76

pherd

0.75

ween

0.74

lished

0.72

qqa

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.