INDEX

Explanations

technical and programming-related terms, especially in the context of Java and mathematical utilities

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

`rhs.repeatedDouble`

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

webElementXpaths

-0.48

InstrumentedTest

-0.47

 betweenstory

-0.46

-------------</

-0.43

styleable

-0.42

 AppCompatTheme

-0.38

OrWhiteSpace

-0.38

 Sett

-0.36

 Repe

-0.35

 Employ

-0.35

POSITIVE LOGITS

 Chwiliwch

0.57

tagHelperRunner

0.56

 cherchés

0.52

AddTagHelper

0.51

 Widerspruch

0.49

 ***!

0.49

WriteBarrier

0.49

друг

0.47

الحياه

0.44

 ligiloj

0.44

Activations Density 0.008%

technical and programming-related terms, especially in the context of Java and mathematical utilities

`rhs.repeatedDouble`

No Comments

No Known Activations