INDEX

Explanations

terms related to models or new concepts in programming or technology

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

end of programming constructs

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

 Bürgermeister

-0.29

 թվական

-0.29

ély

-0.27

 isNameExpr

-0.26

 continúas

-0.26

 stick

-0.26

bleven

-0.26

 taşı

-0.26

 pié

-0.26

PM

-0.25

POSITIVE LOGITS

httphttps

1.03

 AssemblyVersion

0.91

StructEnd

0.88

ngdoc

0.84

 AssemblyProduct

0.84

帖最后由

0.82

 "..\..\

0.79

 Савезне

0.76

onViewCreated

0.75

VersionUID

0.75

Activations Density 0.000%

terms related to models or new concepts in programming or technology

end of programming constructs

No Comments

No Known Activations