INDEX

Explanations

programming-related keywords and expressions linked to function declarations and data handling

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

code formatting

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

المناصب

-0.44

InteropServices

-0.42

 chimie

-0.42

 Antoni

-0.40

__((

-0.40

makeatletter

-0.40

decin

-0.40

.*")]

-0.39

 Meksiku

-0.39

 Army

-0.39

POSITIVE LOGITS

تقاوى

0.44

exitRule

0.43

protoimpl

0.42

EDEFAULT

0.38

(!__

0.35

艋

0.35

GOTREF

0.35

CodeAttribute

0.35

HasAnnotation

0.34

Activations Density 0.027%

programming-related keywords and expressions linked to function declarations and data handling

code formatting

No Comments

No Known Activations