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

uries

-0.71

ingers

-0.68

TO

-0.66

ritten

-0.66

recated

-0.66

aurus

-0.65

entimes

-0.64

ocate

-0.64

inkle

-0.63

prints

-0.63

POSITIVE LOGITS

nels

0.75

 Peng

0.70

nel

0.68

ÑĤ

0.67

cul

0.63

cium

0.63

aten

0.61

vati

0.61

fman

0.60

quist

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.