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

äºº

-0.73

ubes

-0.67

ists

-0.65

igne

-0.64

 Legions

-0.63

mson

-0.63

ishop

-0.62

 Clan

-0.59

idian

-0.59

XVI

-0.59

POSITIVE LOGITS

 mistake

0.79

£ı

0.73

 fallacy

0.73

apo

0.71

 mismatch

0.70

ĨĴ

0.70

pse

0.69

 transact

0.69

hoe

0.68

yip

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.