INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 scram

-0.79

wcs

-0.78

userc

-0.77

rompt

-0.73

sync

-0.70

tackle

-0.69

Password

-0.69

roxy

-0.68

 resil

-0.68

 distingu

-0.67

POSITIVE LOGITS

 Bett

0.74

 Rats

0.74

 Towers

0.68

oth

0.68

 Winged

0.67

phe

0.67

 Extrem

0.66

 Oaks

0.66

 Astro

0.65

 Reyn

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.