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

 scrimmage

-0.73

iability

-0.67

 Yugoslav

-0.66

division

-0.65

esthetic

-0.64

fw

-0.62

ppo

-0.62

esa

-0.62

deb

-0.62

tto

-0.61

POSITIVE LOGITS

 alone

1.43

 Alone

0.87

uyomi

0.79

shire

0.76

regor

0.75

yip

0.72

rael

0.72

minster

0.70

waukee

0.69

oneliness

0.65

Activations Density 0.000%

No Known Activations

This feature has no known activations.