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

thumbnails

-0.70

Interstitial

-0.69

UGC

-0.69

 unarmed

-0.69

ALLY

-0.68

ç¥ŀ

-0.67

liest

-0.67

FACE

-0.67

lly

-0.67

Known

-0.66

POSITIVE LOGITS

atile

0.69

rase

0.69

orie

0.68

 domestically

0.66

 sche

0.65

 Stafford

0.64

rys

0.63

 recruit

0.63

rums

0.62

 satell

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.