INDEX

Explanations

mentions of individuals and their comments or actions, particularly in the context of events and discussions

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

eventually showed

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

 ujednoznacz

-0.65

styleType

-0.56

 disambiguazione

-0.54

 Roskov

-0.50

webElementXpaths

-0.50

ViewFeatures

-0.48

Derbyniad

-0.47

verwijspagina

-0.46

 الحره

-0.46

 transfieras

-0.46

POSITIVE LOGITS

 eventually

0.45

 itself

0.44

ônicos

0.44

eventually

0.42

yet

0.41

GraphicsUnit

0.41

かわらず

0.40

DockStyle

0.40

 Eventually

0.39

Eventually

0.39

Activations Density 0.036%

mentions of individuals and their comments or actions, particularly in the context of events and discussions

eventually showed

No Comments

No Known Activations