INDEX

Explanations

terms related to study methodologies and potential confounding variables in research

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

comparative, ads, random, errors, graphic, private, links, compensation

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

 Taktlose

-0.49

-0.43

 Préférences

-0.39

inaison

-0.39

Mon

-0.38

ⓧ

-0.38

principalColumn

-0.37

 cả

-0.37

ArrowToggle

-0.37

thereal

-0.36

POSITIVE LOGITS

 kasarigan

0.63

writeField

0.62

reportWebVitals

0.60



0.59

RTLR

0.58

 tartalomajánló

0.58

 ligiloj

0.56

ChildScrollView

0.55

 فريبيس

0.55

 ostavi

0.53

Activations Density 0.140%

terms related to study methodologies and potential confounding variables in research

comparative, ads, random, errors, graphic, private, links, compensation

No Comments

No Known Activations