INDEX

Explanations

adjectives or verbs related to morality or ethics

oai_token-act-pair · gpt-3.5-turbo

words related to harm and protection, particularly in social and legal contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

BuyableInstoreAndOnline

-0.64

<?

-0.59

 reluct

-0.59

 fooled

-0.58

 suspic

-0.57

 Laugh

-0.56

aired

-0.55

 nightmare

-0.55

pload

-0.54

 confir

-0.54

POSITIVE LOGITS

 elsewhere

1.14

or

1.04

 abroad

0.91

 themselves

0.88

 respectively

0.86

.[

0.83

 domestically

0.83

 otherwise

0.81

 overseas

0.81

Activations Density 0.395%

adjectives or verbs related to morality or ethics

words related to harm and protection, particularly in social and legal contexts

No Comments

No Known Activations