INDEX

Explanations

negative values and their numerical context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 35-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_35/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.35.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

Personendaten

-0.47

 Superan

-0.39

two

-0.39

 ModelExpression

-0.38

 zwei

-0.37

sizeCache

-0.36

 BoxDecoration

-0.36

istoitu

-0.35

MethodManager

-0.35

 three

-0.34

POSITIVE LOGITS

thousands

0.66

 thousands

0.65

 hundreds

0.65

 THOUS

0.63

hundreds

0.61

 HUND

0.59

 thousand

0.58

thousand

0.56

Thousands

0.56

 hundred

0.55

Activations Density 1.314%

negative values and their numerical context

No Comments

No Known Activations

negative values and their numerical context

No Comments

No Known Activations