INDEX

Explanations

structural components and attributes related to data management and programming

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

code identifiers and keywords

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

帖最后由

-0.57

AddTagHelper

-0.54

setVerticalGroup

-0.54

Obrázky

-0.51

 Савезне

-0.48

CPtr

-0.48

ImageContext

-0.46

__':

-0.46

رشف

-0.46

transQ

-0.45

POSITIVE LOGITS

AntiForgeryToken

0.43

OGND

0.41

WithIdentifier

0.40

 GenerationType

0.39

 traditionnels

0.38

EClass

0.35

ابراین

0.34

TagMode

0.34

 teda

0.33

備

0.33

Activations Density 0.058%

structural components and attributes related to data management and programming

code identifiers and keywords

No Comments

No Known Activations