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

adesh

-0.81

elaide

-0.73

ieth

-0.72

itals

-0.71

aughs

-0.71

ocious

-0.67

pless

-0.63

obbies

-0.63

acho

-0.62

Ship

-0.62

POSITIVE LOGITS

 Strauss

0.71

 intermedi

0.69

INK

0.68

 Manor

0.66

 Mush

0.65

 Cheney

0.65

TAG

0.64

riks

0.63

jon

0.63

 Bernstein

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.