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

borgh

-0.90

cipled

-0.82

eland

-0.79

pload

-0.78

assic

-0.78

vironment

-0.78

nai

-0.77

dylib

-0.75

wcs

-0.74

EH

-0.73

POSITIVE LOGITS

ures

0.80

 these

0.80

Era

0.68

uring

0.64

these

0.63

ured

0.62

 Hancock

0.61

 Series

0.61

 These

0.61

uren

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.