INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Buzz

-0.72

utical

-0.71

bara

-0.69

ounce

-0.67

 Authorization

-0.65

heit

-0.64

erton

-0.64

ç«

-0.63

uyomi

-0.62

 Search

-0.62

POSITIVE LOGITS

 truce

0.71

icho

0.71

lycer

0.65

 clos

0.64

itars

0.63

 ensu

0.63

 rapist

0.62

FH

0.61

icides

0.61

seless

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.