INDEX

Explanations

references to detectives and detective work

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

detective or investigator

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

Biomass

-0.58

ToBounds

-0.57

 kasarigan

-0.56

 totalSupply

-0.55

 tallest

-0.54

 présidentielle

-0.54

 rígida

-0.53

 vogliono

-0.53

Obesity

-0.52

maßen

-0.52

POSITIVE LOGITS

 detective

1.08

 Detective

0.94

 detectives

0.93

detective

0.84

Detective

0.81

 Dete

0.71

 DETECT

0.67

 investigator

0.65

 dete

0.61

探偵

0.60

Activations Density 0.006%

references to detectives and detective work

detective or investigator

No Comments

No Known Activations