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

worth

-0.64

tul

-0.62

 curtains

-0.62

owered

-0.62

 curtain

-0.62

 inner

-0.61

scl

-0.61

anism

-0.60

ora

-0.60

 cryptoc

-0.58

POSITIVE LOGITS

actor

0.75

oux

0.68

 Gladiator

0.65

TA

0.64

Train

0.63

advert

0.62

Hero

0.61

 Leilan

0.61

ãĥĢ

0.61

 Wilderness

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.