INDEX

Explanations

extreme qualities or attributes described in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 12-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_12/width_16k/average_l0_130

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.12.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

 الرياضيه

-0.71

цездатний

-0.50

 تانيه

-0.45

Geografía

-0.44

 ContentValues

-0.43

ništvo

-0.40

mäh

-0.39

#+#

-0.38

 محفوظة

-0.38

WindowConstants

-0.38

POSITIVE LOGITS

 extreme

0.65

 Extreme

0.63

extreme

0.63

 extrême

0.61

Extreme

0.60

 extremo

0.60

 Extremely

0.58

Extremely

0.58

 extremadamente

0.56

 ultra

0.54

Activations Density 0.958%

extreme qualities or attributes described in various contexts

No Comments

No Known Activations