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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

CVE

-0.74

eers

-0.72

 suspic

-0.68

eer

-0.66

CLASSIFIED

-0.64

esthes

-0.62

Ñģ

-0.62

omical

-0.61

Sensor

-0.61

RFC

-0.61

POSITIVE LOGITS

 directly

1.09

to

0.92

 instead

0.72

izont

0.69

 indirectly

0.63

inately

0.63

 embed

0.62

 alone

0.61

azon

0.61

 efficiently

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.