INDEX

Explanations

statements addressing or directing actions towards the reader

oai_token-act-pair · gpt-3.5-turbo

phrases that directly address or involve the reader

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

Els

-0.85

rone

-0.70

 Ambro

-0.69

 Jagu

-0.65

 Mata

-0.64

AME

-0.64

uca

-0.63

erella

-0.63

onto

-0.61

 âĵĺ

-0.61

POSITIVE LOGITS

 refres

0.70

 reader

0.66

yout

0.65

ibrarian

0.64

 disclaim

0.62

 amended

0.61

atre

0.60

 critic

0.60

 recl

0.60

ource

0.60

Activations Density 0.277%

statements addressing or directing actions towards the reader

phrases that directly address or involve the reader

No Comments

No Known Activations