INDEX

Explanations

content with explicit or graphic content

oai_token-act-pair · gpt-3.5-turbo

references to social and cultural commentary, often with informal and humorous undertones

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

isSpecialOrderable

-0.73

asury

-0.69

Reward

-0.67

 safegu

-0.66

Import

-0.66

åĬ

-0.66

 qualitative

-0.65

Clear

-0.65

¿½

-0.65

 Regulatory

-0.63

POSITIVE LOGITS

lol

1.41

 haha

1.39

;)

1.31

LOL

1.31

?!

1.30

!!!!

1.27

???

1.23

!!!!!

1.23

!!!

1.22

 shit

1.22

Activations Density 0.618%

content with explicit or graphic content

references to social and cultural commentary, often with informal and humorous undertones

No Comments

No Known Activations