INDEX

Explanations

evaluative adjectives and descriptors reflecting quality or effectiveness

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

Adjective followed by a positive-tone noun

np_acts-logits-general · gemini-2.0-flash

evaluative adjectives followed by nouns

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 للاسماء

-0.64



-0.61

Ligações

-0.60

 Taktlose

-0.60

 utafitiHapana

-0.59

 saites

-0.58

oneofs

-0.58

AISSEE

-0.56

ToListAsync

-0.56

сылкі

-0.55

POSITIVE LOGITS

 kind

0.46

way

0.43

openhauer

0.39

 part

0.39

 platform

0.38

떤

0.37

 guma

0.37

 component

0.36

存于互联网档案馆

0.36

 ترین

0.36

Activations Density 0.302%

evaluative adjectives and descriptors reflecting quality or effectiveness

Adjective followed by a positive-tone noun

evaluative adjectives followed by nouns

No Comments

No Known Activations