INDEX

Explanations

words related to opinions or editorials

oai_token-act-pair · gpt-3.5-turbo

terms related to opioid substances and discussions surrounding them

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ĸļ

-0.66

 confidentiality

-0.64

omez

-0.63

 srfAttach

-0.62

infeld

-0.59

Hik

-0.58

WARE

-0.56

inki

-0.56

POW

-0.56

 Cobb

-0.56

POSITIVE LOGITS

lisher

0.99

monary

0.97

urized

0.75

leck

0.74

essler

0.74

lishing

0.71

rencies

0.71

ciation

0.69

blem

0.68

elines

0.68

Activations Density 0.105%

words related to opinions or editorials

terms related to opioid substances and discussions surrounding them

No Comments

No Known Activations