INDEX

Explanations

adjectives that denote quality, effectiveness, or value in various contexts

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

creation of positive outcomes

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

 paciencia

-0.44

 rimasto

-0.42

 resté

-0.42

 Zähne

-0.41

 dureza

-0.41

 hayas

-0.39

 chiunque

-0.39

ContextCompat

-0.39

 creencias

-0.38

 antiga

-0.38

POSITIVE LOGITS

CreateModel

0.68

CreateMap

0.65

 create

0.59

create

0.59

 createNew

0.57

tvguidetime

0.57

 skapa

0.54

creating

0.54

createCell

0.53

 creating

0.53

Activations Density 0.091%

adjectives that denote quality, effectiveness, or value in various contexts

creation of positive outcomes

No Comments

No Known Activations