INDEX

Explanations

the word "See" at varying levels of activation

oai_token-act-pair · gpt-3.5-turbo

phrases related to viewing content or information

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

naires

-0.64

naire

-0.61

angan

-0.58

ngth

-0.58

othal

-0.56

alky

-0.54

 hurd

-0.54

elf

-0.54

iod

-0.54

posal

-0.54

POSITIVE LOGITS

 Sample

0.59

 Cancel

0.54

 Sketch

0.54

Property

0.53

Widget

0.53

 whats

0.53

 Entreprene

0.53

Vu

0.52

 Shooting

0.52

 previews

0.52

Activations Density 0.013%

the word "See" at varying levels of activation

phrases related to viewing content or information

No Comments

No Known Activations