INDEX

Explanations

articles and demonstrative pronouns indicating specific entities or items

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

the word "the" (appears to have high activation values near "the" across multiple contexts).

oai_token-act-pair · claude-3-5-haiku-20241022 Triggered by @neilrathi

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 следует

-0.52

应当

-0.44

 estrecha

-0.43

 współpracy

-0.42

 seno

-0.41

 należy

-0.41

 also

-0.40

 слід

-0.40

 samband

-0.39

 yılında

-0.39

POSITIVE LOGITS

 mxArray

0.63

Diweddarwch

0.63

 kaarangay

0.61

 estekak

0.60

jsxFileName

0.59

 nonUne

0.58

 linkovi

0.57

 Infórmanos

0.57

 архивлан

0.55

 sumpay

0.53

Activations Density 0.228%

articles and demonstrative pronouns indicating specific entities or items

the word "the" (appears to have high activation values near "the" across multiple contexts).

No Comments

No Known Activations