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

 itiner

-0.75

 pedestrians

-0.62

 Haste

-0.62

 Ally

-0.61

Hom

-0.60

 Koen

-0.60

 Merry

-0.60

 invari

-0.59

ified

-0.58

 BELOW

-0.58

POSITIVE LOGITS

Ã£

0.91

escription

0.84

amiya

0.84

idium

0.80

rish

0.77

hooting

0.76

cient

0.76

ammers

0.74

asa

0.73

aspers

0.73

Activations Density 0.000%

No Known Activations

This feature has no known activations.