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

tackle

-0.94

upuncture

-0.77

reath

-0.76

ervative

-0.71

etooth

-0.69

undreds

-0.68

inian

-0.68

isine

-0.66

hari

-0.66

pet

-0.66

POSITIVE LOGITS

 unpop

0.66

Lag

0.65

 Azerb

0.65

 Rosenberg

0.65

Kru

0.64

 Levy

0.64

 pall

0.63

 Palest

0.63

 Narr

0.62

 unification

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.