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

 exerc

-0.79

 seiz

-0.72

ogl

-0.71

xual

-0.71

reads

-0.68

igne

-0.66

 Emblem

-0.66

ovo

-0.66

iciency

-0.65

Fn

-0.64

POSITIVE LOGITS

CHAT

0.67

aciously

0.65

Tell

0.65

 smugglers

0.64

Southern

0.63

çİĭ

0.61

Wan

0.59

Delta

0.59

BTC

0.58

 overd

0.58

Activations Density 0.000%

No Known Activations

This feature has no known activations.