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

ahu

-0.85

acebook

-0.79

ember

-0.79

ormal

-0.75

cher

-0.75

ocument

-0.74

abul

-0.72

vity

-0.71

ilogy

-0.71

aturday

-0.67

POSITIVE LOGITS

 Hits

0.70

"âĢ¦

0.70

 Breach

0.66

huh

0.63

asel

0.61

EStream

0.60

heard

0.60

 Guant

0.59

dod

0.59

 joints

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.