INDEX

Explanations

programming constructs and structures

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

conditional statements or definitions

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

 betweenstory

-0.63

ſelf

-0.56

Ӕ

-0.52

WebElementEntity

-0.51

 purpoſe

-0.48

 NUKAT

-0.48

 Signalez

-0.48

 verſ

-0.47

AndEndTag

-0.47

 dezelve

-0.46

POSITIVE LOGITS

complexContent

0.49

endmodule

0.48

 bear

0.43

 clubhouse

0.43

 define

0.42

ubs

0.41

 function

0.41

 start

0.40

 defined

0.40

createStatement

0.40

Activations Density 1.117%

programming constructs and structures

conditional statements or definitions

No Comments

No Known Activations