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

forth

-0.76

ctuary

-0.74

 Sever

-0.70

minist

-0.70

ERC

-0.69

jab

-0.66

ech

-0.65

abus

-0.65

ITH

-0.64

ILY

-0.64

POSITIVE LOGITS

 resembled

0.86

 resemble

0.76

 insign

0.74

 emulate

0.74

ogun

0.73

ggles

0.73

 aspire

0.72

 differentiate

0.72

 populate

0.72

ailability

0.71

Activations Density 0.000%

No Known Activations

This feature has no known activations.