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

 Garland

-0.66

 Shell

-0.66

 Bullets

-0.64

 Generations

-0.64

 Rounds

-0.64

 Salam

-0.63

 Impossible

-0.62

dark

-0.60

 Adventures

-0.60

 motions

-0.59

POSITIVE LOGITS

0.90

ictive

0.80

cedes

0.79

ãĤ®

0.79

catentry

0.77

fet

0.72

abin

0.71

oslav

0.70

lect

0.70

consumer

0.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.