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

theless

-0.80

nect

-0.68

 scrolls

-0.62

noxious

-0.61

visual

-0.61

hound

-0.60

 stunt

-0.59

verages

-0.58

 apiece

-0.58

cott

-0.58

POSITIVE LOGITS

am

0.99

enges

0.65

EM

0.65

cules

0.64

doms

0.62

Rez

0.62

ei

0.62

uminati

0.62

roxy

0.59

ublic

0.59

Activations Density 0.000%

No Known Activations

This feature has no known activations.