INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ovie

-0.83

vernment

-0.78

mire

-0.75

iqueness

-0.72

âĹ¼

-0.70

 Locked

-0.69

Owl

-0.67

locked

-0.67

uyomi

-0.66

================

-0.65

POSITIVE LOGITS

essing

0.71

 Course

0.66

 whence

0.65

atan

0.65

ect

0.64

itters

0.62

ences

0.62

auri

0.62

Â¶

0.62

ossom

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.