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

ipel

-0.79

iliate

-0.73

yip

-0.64

neau

-0.63

 coordinated

-0.63

export

-0.63

 cataly

-0.63

KA

-0.61

 uncond

-0.61

 unchecked

-0.60

POSITIVE LOGITS

 words

0.92

 Creed

0.74

 Words

0.73

 Matter

0.69

gravity

0.65

 Streets

0.64

Square

0.64

 Word

0.64

mort

0.63

©¶æ¥µ

0.62

Activations Density 0.000%

No Known Activations

This feature has no known activations.