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

estro

-0.74

uana

-0.72

zone

-0.72

ursday

-0.71

eless

-0.70

ochet

-0.69

anium

-0.68

mine

-0.67

ulhu

-0.66

apsed

-0.65

POSITIVE LOGITS

Īè

0.84

Ô

0.80

Reply

0.73

 Merrill

0.72

 Bere

0.70

 Verse

0.70

Ple

0.70

 Merry

0.69

 Boll

0.68

Lip

0.67

Activations Density 0.000%

No Known Activations

This feature has no known activations.