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

hement

-0.80

quart

-0.78

 Pens

-0.75

Thirty

-0.71

Benz

-0.70

pend

-0.69

illes

-0.66

cellence

-0.66

 miscon

-0.65

wid

-0.65

POSITIVE LOGITS

eria

0.81

ched

0.71

è¦

0.65

auna

0.65

enda

0.65

aves

0.65

aved

0.64

ori

0.64

ice

0.64

aily

0.63

Activations Density 0.000%

No Known Activations

This feature has no known activations.