INDEX

Explanations

terms related to data processing and configuration settings

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

C++ testing utilities

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

 problematic

-0.45

 createState

-0.44

 dolayı

-0.42

 navidad

-0.41

RRect

-0.40

 Whilst

-0.40

 towards

-0.39

ϖ

-0.39

 realise

-0.39

 problems

-0.39

POSITIVE LOGITS

ſelf

0.51

PhysRevD

0.50

 unſ

0.48

zun

0.47

 ſind

0.47

$_(

0.46

RTHOOK

0.46

Eſ

0.46

 coö

0.46

 الرياضيه

0.46

Activations Density 0.062%

terms related to data processing and configuration settings

C++ testing utilities

No Comments

No Known Activations