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

onduct

-0.71

 Enhancement

-0.71

OA

-0.68

atically

-0.66

 Nieto

-0.63

"}

-0.63

henko

-0.62

lishes

-0.57

ileaks

-0.57

kefeller

-0.57

POSITIVE LOGITS

volume

0.69

ãĥ´

0.66

wings

0.66

thinkable

0.64

voy

0.64

¬¼

0.62

ighters

0.62

Kal

0.62

]+

0.61

æĪ¦

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.