INDEX

Explanations

people and entities related to academia and university, especially with a focus on Sacramento and nuclear-related terms

oai_token-act-pair · gpt-3.5-turbo

Triggered by @bot

New Auto-Interp

Configuration

jbloom/Gemma-2b-Residual-Stream-SAEs/gemma_2b_blocks.0.hook_resid_post_16384_anthropic

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

HuggingFaceFW/fineweb

Features

16,384

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

HuggingFaceFW/fineweb

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 endow

-0.78

 intersper

-0.69

<bos>

-0.69

 gespannt

-0.67

 superintend

-0.66

towania

-0.64

acyjny

-0.63

 plundered

-0.63

 heighten

-0.63

 ascribe

-0.62

POSITIVE LOGITS

kac

1.56

 maksi

1.53

Sac

1.46

 panik

1.44

Bac

1.34

 silikon

1.31

 makro

1.30

sac

1.29

Lac

1.28

 antik

1.28

Activations Density 0.281%

people and entities related to academia and university, especially with a focus on Sacramento and nuclear-related terms

No Known Activations