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

 caption

-0.82

vernment

-0.76

aution

-0.72

 subtitles

-0.72

 graffiti

-0.70

 subtitle

-0.69

displayText

-0.69

 Lauder

-0.69

 Manz

-0.68

 signage

-0.68

POSITIVE LOGITS

execute

0.72

leigh

0.72

Delete

0.71

pher

0.64

METHOD

0.63

Higher

0.62

qual

0.61

cot

0.61

pm

0.60

Eastern

0.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.