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

account

-0.77

 Divide

-0.74

omsday

-0.73

estate

-0.72

matter

-0.72

Requ

-0.71

credit

-0.70

chell

-0.70

limits

-0.68

 Limit

-0.67

POSITIVE LOGITS

 bush

0.82

oller

0.81

 tuber

0.69

Ely

0.65

 VIDE

0.64

 Cros

0.63

 plush

0.63

ãĤ©

0.63

lication

0.61

udeb

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.