INDEX

Explanations

phrases indicating action or determination

oai_token-act-pair · gpt-3.5-turbo

phrases related to actions taken or responsibilities fulfilled

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Yards

-0.66

 Cong

-0.63

 Returning

-0.62

uttering

-0.61

 Vaugh

-0.61

 Hills

-0.59

 Motor

-0.57

end

-0.57

 inacc

-0.56

haus

-0.55

POSITIVE LOGITS

soever

1.20

 happens

0.99

 happened

0.85

 kinds

0.85

 sorts

0.79

Downloadha

0.78

 else

0.78

til

0.75

 mattered

0.73

DragonMagazine

0.72

Activations Density 0.113%

phrases indicating action or determination

phrases related to actions taken or responsibilities fulfilled

No Comments

No Known Activations