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

 mosqu

-0.74

nels

-0.73

ileged

-0.69

 Emirates

-0.67

 friendly

-0.67

 slightest

-0.67

 connections

-0.67

TAG

-0.67

ties

-0.66

 onlook

-0.65

POSITIVE LOGITS

amp

2.17

amps

1.22

amping

1.04

shire

0.86

oldown

0.83

aird

0.77

ende

0.77

cott

0.76

gow

0.73

cot

0.72

Activations Density 0.000%

No Known Activations

This feature has no known activations.