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

wana

-0.82

owan

-0.77

ilial

-0.73

 Petersen

-0.67

agall

-0.65

 Willis

-0.63

IAL

-0.62

IFIED

-0.62

aho

-0.62

emort

-0.61

POSITIVE LOGITS

ocalypse

0.82

ocalyptic

0.67

genre

0.66

cession

0.64

 whis

0.62

ombies

0.62

beh

0.61

Mania

0.61

 compressor

0.61

///

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.