INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 mechanic

-0.75

 visibly

-0.71

 cris

-0.69

 redist

-0.65

 whip

-0.64

 Mechan

-0.64

 charismatic

-0.64

 mechanically

-0.63

Aut

-0.62

 lightly

-0.62

POSITIVE LOGITS

MpServer

0.87

apeake

0.83

rees

0.82

asus

0.79

xon

0.77

ilater

0.77

ongyang

0.76

eus

0.74

oplan

0.73

à¼

0.73

Activations Density 0.000%

No Known Activations

This feature has no known activations.