INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Flask

-0.74

yers

-0.74

 redes

-0.73

yip

-0.71

anchester

-0.69

 Bundes

-0.65

 SHAR

-0.65

erto

-0.64

llah

-0.64

 ACTIONS

-0.62

POSITIVE LOGITS

okemon

0.80

merce

0.66

axy

0.62

opolis

0.61

Moy

0.61

rama

0.58

>]

0.57

trace

0.56

 Archdemon

0.56

 manure

0.56

Activations Density 0.000%

No Known Activations

This feature has no known activations.