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

 penetrating

-0.75

ATCH

-0.70

onz

-0.68

ulhu

-0.66

 somewhere

-0.65

KI

-0.65

ancouver

-0.64

Pear

-0.62

 guiActive

-0.61

HO

-0.60

POSITIVE LOGITS

shire

0.73

theless

0.70

adian

0.70

iasis

0.70

thal

0.70

 ç¥ŀ

0.65

 Fold

0.63

 Fixes

0.62

eared

0.62

olit

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.