INDEX

Explanations

structures related to lists and categorization

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

the winners are

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

 kaynağından

-0.59

 ***!

-0.58

writeFieldEnd

-0.55

MIDDLEWARE

-0.54

 Italijani

-0.53

styleType

-0.53

 kasarigan

-0.52

########.

-0.50

 emailAlready

-0.50

 GALERÍA

-0.50

POSITIVE LOGITS

ubio

0.38

iredo

0.38

所有

0.37

 propOrder

0.37

anele

0.35

bio

0.35

0.34

部

0.34

 list

0.34

InputTagHelper

0.34

Activations Density 0.032%

structures related to lists and categorization

the winners are

No Comments

No Known Activations

structures related to lists and categorization

the winners are

No Comments

No Known Activations