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

mie

-0.87

enges

-0.82

senal

-0.81

encia

-0.79

tesy

-0.79

 Pixie

-0.77

veyard

-0.74

ruction

-0.72

ACY

-0.71

amia

-0.70

POSITIVE LOGITS

obin

0.73

quished

0.67

 treaties

0.64

 barg

0.64

 fitt

0.63

 entitle

0.62

 heroine

0.62

 commanding

0.62

istas

0.62

 wonder

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.