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

NPR

-0.62

JD

-0.61

 Annotations

-0.60

SI

-0.60

UU

-0.60

 arsen

-0.59

 Sinai

-0.59

 Garcia

-0.59

 Shutterstock

-0.58

 cheers

-0.58

POSITIVE LOGITS

mins

0.76

cially

0.74

ufact

0.71

 Nurs

0.71

vell

0.69

heit

0.68

pieces

0.66

anium

0.65

 swearing

0.65

.''.

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.