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

oster

-0.70

xia

-0.68

stadt

-0.60

 bases

-0.59

ģ«

-0.59

 mentally

-0.59

AAAA

-0.57

bud

-0.57

 wrestler

-0.57

eties

-0.57

POSITIVE LOGITS

 upon

1.36

upon

1.00

 Upon

0.82

pell

0.73

 Blessing

0.69

Upon

0.64

illet

0.61

nces

0.61

vous

0.61

semble

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.