INDEX

Explanations

proper nouns, specifically names and titles of people or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 36-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_36/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.36.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

RegressionTest

-0.61

rungsseite

-0.57

vician

-0.49

AccessorTable

-0.44

 ComVisible

-0.43

enumii

-0.43

Derbyniad

-0.41

postsleuth

-0.37

contentLoaded

-0.37

Xna

-0.37

POSITIVE LOGITS

 last

3.98

last

3.30

 último

3.02

Last

3.02

 Last

2.97

 última

2.92

 terakhir

2.91

 LAST

2.88

 dernier

2.86

 dernière

2.86

Activations Density 1.301%

proper nouns, specifically names and titles of people or entities

No Comments

No Known Activations