INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

dden

-0.77

alach

-0.74

 acquaint

-0.70

 Reach

-0.68

adian

-0.68

uate

-0.64

cised

-0.63

FN

-0.63

ohan

-0.62

 Lash

-0.62

POSITIVE LOGITS

ILLE

0.76

ession

0.66

 UEFA

0.64

CLR

0.63

 Wenger

0.62

NES

0.62

iggins

0.61

Textures

0.60

 extracts

0.60

 racks

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.