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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

arter

-0.73

ilda

-0.68

é¾

-0.67

Roaming

-0.66

ãģ®å®

-0.66

Role

-0.65

 Stealth

-0.65

afe

-0.65

independence

-0.64

 disband

-0.62

POSITIVE LOGITS

abilia

0.75

 contrace

0.72

acers

0.71

itia

0.68

 grabs

0.68

ttes

0.66

alian

0.62

osphere

0.61

 Unsure

0.61

bable

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.