INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

itage

-0.86

opian

-0.84

orescence

-0.84

axy

-0.81

emis

-0.80

 earthqu

-0.77

arser

-0.76

 tradem

-0.76

 queues

-0.74

 busiest

-0.73

POSITIVE LOGITS

Story

1.08

Breaking

1.02

Spread

1.00

Rating

1.00

CLOSE

0.99

ÃĹ

0.97

Still

0.96

Arsenal

0.96

HOU

0.96

Overview

0.94

Activations Density 0.000%

No Known Activations

This feature has no known activations.