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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

abulary

-0.72

 handcuffs

-0.65

ierrez

-0.65

 exported

-0.65

 watered

-0.64

 largeDownload

-0.64

ccording

-0.62

 Upload

-0.61

 deported

-0.61

uthor

-0.60

POSITIVE LOGITS

 Whale

0.78

side

0.77

lest

0.76

 Mellon

0.75

 Pulitzer

0.71

edin

0.71

ero

0.65

orum

0.65

 Danger

0.64

halla

0.64

Activations Density 0.000%

No Known Activations

This feature has no known activations.