INDEX

Explanations

code-related elements, particularly in the context of programming structure

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

user turn marker

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

httphttps

-0.40

exitRule

-0.40

ніципалі

-0.38

expandindo

-0.34

SequentialGroup

-0.34

 незавершена

-0.32

Externe

-0.32

 שוליים

-0.31

exploration

-0.31

 autorytatywna

-0.30

POSITIVE LOGITS

IntoConstraints

0.63

RectangleBorder

0.56

ViewInit

0.52

 compactness

0.50

astify

0.47

 ChromeDriver

0.47

NameInMap

0.46

InputBorder

0.46

JNIEnv

0.45

 AssemblyCompany

0.44

Activations Density 0.009%

code-related elements, particularly in the context of programming structure

user turn marker

No Comments

No Known Activations