INDEX

Explanations

terms related to positive feelings and experiences

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

pleasant

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

 NSCoder

-0.65

 المعيارى

-0.59

Autoritní

-0.57

 Suivez

-0.57

Välislingid

-0.56

 wireType

-0.55

DatabaseError

-0.54

 motorbike

-0.53

 virginity

-0.53

โคร

-0.52

POSITIVE LOGITS

 Pleasant

0.88

Pleasant

0.85

pleasant

0.83

 pleasant

0.81

 agradable

0.61

 unpleasant

0.60

 angene

0.56

 прият

0.56

Nice

0.55

 agréable

0.53

Activations Density 0.002%

terms related to positive feelings and experiences

pleasant

No Comments

No Known Activations