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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

imilation

-0.82

eatures

-0.80

idity

-0.79

ente

-0.77

arius

-0.76

ylum

-0.72

itia

-0.71

auga

-0.71

Jah

-0.70

Ire

-0.70

POSITIVE LOGITS

seed

0.75

Fax

0.68

 PAGE

0.68

RFC

0.64

EVA

0.64

Wood

0.64

Tact

0.63

CF

0.63

Flight

0.62

MIT

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.