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

conom

-0.71

 Airl

-0.66

atten

-0.63

ĸļ

-0.61

handle

-0.60

 fashioned

-0.60

dimensional

-0.60

Disney

-0.60

knit

-0.59

seek

-0.58

POSITIVE LOGITS

 Cosponsors

0.95

iencies

0.83

chnology

0.70

ospels

0.70

cible

0.70

asting

0.67

ucc

0.66

orsi

0.66

ync

0.65

ription

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.