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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

aqu

-0.84

arnaev

-0.83

perty

-0.82

eport

-0.81

ighth

-0.80

ategory

-0.76

ancial

-0.72

alez

-0.69

uberty

-0.69

lore

-0.69

POSITIVE LOGITS

BU

0.78

 respons

0.75

 Bihar

0.73

 theirs

0.68

Kit

0.65

 Instr

0.64

Narr

0.63

 spons

0.63

 Participant

0.62

 ours

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.