INDEX

Explanations

patterns related to coding or programming constructs

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

programming keywords

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

 AssemblyCulture

-0.77

 queſta

-0.77

windowFixed

-0.76

SBATCH

-0.76

transQ

-0.73

 ivelany

-0.73

parsedMessage

-0.73

 '\\;'

-0.71

########.

-0.69

 виправивши

-0.68

POSITIVE LOGITS

His

0.39

0.37

 major

0.35

 Major

0.34

↵

0.33

0.32

 matéria

0.32

major

0.31

Major

0.31

<strong>

0.31

Activations Density 0.018%

patterns related to coding or programming constructs

programming keywords

No Comments

No Known Activations