INDEX

Explanations

terms related to censorship and control, especially in the context of offensive language

oai_token-act-pair · gpt-3.5-turbo

words and phrases related to assistance or prevention

oai_token-act-pair · gpt-4-turbo

discussions related to the prevention of harmful or offensive actions and terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

uploads

-0.50

 largeDownload

-0.49

 Morty

-0.49

pse

-0.46

 Hide

-0.42

 nutshell

-0.40

 âĶľ

-0.40

pmwiki

-0.40

 Brewers

-0.40

 Fine

-0.40

POSITIVE LOGITS

livion

0.50

 footing

0.50

izont

0.45

omever

0.44

$.

0.44

burse

0.44

 someday

0.43

ictionary

0.43

ãģ¾

0.43

poke

0.43

Activations Density 6.560%

terms related to censorship and control, especially in the context of offensive language

words and phrases related to assistance or prevention

discussions related to the prevention of harmful or offensive actions and terms

No Comments

No Known Activations