INDEX

Explanations

phrases indicating breaking down or analyzing something

oai_token-act-pair · gpt-3.5-turbo

phrases related to deconstruction or analysis

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

istas

-0.68

 Gunn

-0.64

Lv

-0.62

 Jinn

-0.62

ught

-0.62

 Promise

-0.62

mong

-0.62

jab

-0.60

 hypert

-0.59

inen

-0.59

POSITIVE LOGITS

sheets

0.80

 carbohydrates

0.76

taining

0.75

casts

0.74

 barriers

0.74

CTR

0.73

grading

0.72

alsa

0.72

shit

0.70

inately

0.70

Activations Density 0.025%

phrases indicating breaking down or analyzing something

phrases related to deconstruction or analysis

No Comments

No Known Activations