INDEX

Explanations

proper nouns, specifically names of places and companies

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

company locations

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

 behaviours

-0.34

 behaviour

-0.34

 komen

-0.33

seins

-0.33

 Gedächt

-0.32

╣

-0.32

Spoljašnje

-0.32

 akár

-0.31

UnusedPrivate

-0.31

 članak

-0.30

POSITIVE LOGITS

 headquartered

0.77

 Houſe

0.71

الحياه

0.71

quartered

0.67

 headquarters

0.64

GEBURTS

0.61

总部

0.59

 houſe

0.58

脚注の使い方

0.57

MessageState

0.57

Activations Density 0.036%

proper nouns, specifically names of places and companies

company locations

No Comments

No Known Activations