INDEX

Explanations

instances of code-related keywords and programming constructs

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

it either

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

Negative Logits

Hochspringen

-0.49

SequentialGroup

-0.46

tonode

-0.45

 >=",

-0.40

 verdad

-0.37

 Mitteln

-0.36

Conclusión

-0.36

 tartalomajánló

-0.35

 PyLong

-0.35

Nationalité

-0.34

POSITIVE LOGITS

 disambiguazione

0.60

insics

0.55

asteroide

0.54

 invokingState

0.52

 ujednoznacz

0.52

oredCriteria

0.50

 setw

0.49

 المعيارى

0.48

كويكب

0.48

uxxxx

0.48

Activations Density 0.150%

instances of code-related keywords and programming constructs

it either

No Known Activations