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

tein

-0.84

horn

-0.82

vironment

-0.78

pi

-0.77

unders

-0.77

hemy

-0.73

soDeliveryDate

-0.73

raf

-0.72

rir

-0.72

auld

-0.71

POSITIVE LOGITS

NCT

0.82

 monop

0.62

ret

0.61

SAM

0.59

 attendant

0.58

 Mazda

0.58

 Hemp

0.57

iph

0.56

 McCoy

0.56

 taxpayers

0.56

Activations Density 0.000%

No Known Activations

This feature has no known activations.