INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

iversary

-0.75

luaj

-0.67

 Schne

-0.64

SN

-0.64

ctuary

-0.62

apest

-0.62

 ãĤµ

-0.61

 Afgh

-0.60

shine

-0.60

atom

-0.59

POSITIVE LOGITS

awei

0.73

 proportions

0.71

igslist

0.70

outube

0.70

ongh

0.70

rists

0.69

alf

0.66

pec

0.62

olves

0.62

cellaneous

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.