INDEX

Explanations

terms related to "bi" or "bipartisan" concepts, indicating a focus on bipartisanship or duality in subjects

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

bi words or Bi element

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

UserScript

-0.55

 honneur

-0.55

 preparación

-0.55

 disambiguazione

-0.54

 vänt

-0.52

 humanidad

-0.52

 Generated

-0.51

unicaciones

-0.51

 esternos

-0.50

setLength

-0.50

POSITIVE LOGITS

Bi

0.88

Bi

0.81

bi

0.80

Bim

0.78

BI

0.74

Bim

0.74

Bif

0.69

Bif

0.68

bi

0.67

bim

0.66

Activations Density 0.024%

terms related to "bi" or "bipartisan" concepts, indicating a focus on bipartisanship or duality in subjects

bi words or Bi element

No Comments

No Known Activations