INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Tsukuyomi

-0.78

Nak

-0.66

else

-0.65

 otherwise

-0.64

endif

-0.63

 hereafter

-0.63

 else

-0.62

Nou

-0.61

 settlement

-0.60

ppo

-0.58

POSITIVE LOGITS

ularity

0.73

inki

0.71

ragon

0.71

ushima

0.70

milo

0.69

tymology

0.69

conduct

0.68

urized

0.68

Minecraft

0.68

ierrez

0.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.