INDEX

Explanations

email addresses or related components

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

line starts with "user"

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

 proti

-0.31

 plainte

-0.27

 Mathf

-0.27

 murs

-0.26

 Lihat

-0.26

 hård

-0.25

UINTN

-0.25

IsOptional

-0.24

reportWebVitals

-0.24

 fija

-0.24

POSITIVE LOGITS

帖最后由

0.71

OGND

0.67

setVerticalGroup

0.65

 nakalista

0.59

QMetaType

0.59

enumii

0.58

kháu

0.57

ImageContext

0.57

GEBURTSDATUM

0.56

المناصب

0.55

Activations Density 0.000%

email addresses or related components

line starts with "user"

No Comments

No Known Activations