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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

NER

-0.80

pmwiki

-0.72

sers

-0.70

 EDITION

-0.65

Reviewer

-0.65

rences

-0.64

ASED

-0.63

buster

-0.63

ALLY

-0.63

 remake

-0.62

POSITIVE LOGITS

pires

0.80

bestos

0.78

well

0.77

pired

0.74

criptions

0.72

icho

0.70

ifice

0.70

par

0.68

ovych

0.67

hy

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.