INDEX

Explanations

specific markers or tokens indicating the start of content

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

sizeCache

-0.49

MathML

-0.48

 Photocase

-0.47

 culturelles

-0.43

dessous

-0.43

 financières

-0.41

collation

-0.41

 affich

-0.40

因为它

-0.40

Springfield

-0.40

POSITIVE LOGITS

 aload

0.61

 iVar

0.57

="@+

0.52

 CURLOPT

0.52

enumi

0.51

GOTREF

0.51

 dumne

0.47

adox

0.47

 Benjamin

0.46

Activations Density 0.005%

specific markers or tokens indicating the start of content

No Comments

No Known Activations