INDEX

Explanations

references to personal experiences and identities

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

my name

np_acts-logits-general · gemini-2.5-flash-lite

When text is a chat/forum/IRC speaker label (a username or nick often shown in angle brackets or as a line author), this neuron strongly activates.

oai_token-act-pair · gpt-5-mini Triggered by @vetterc0

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

 couleurs

-0.43

 consommateurs

-0.42

原始内容存档于

-0.41

 gratuitement

-0.40

 remboursement

-0.40

dedores

-0.40

 pelatihan

-0.39

 training

-0.39

 aprendizagem

-0.37

 kolory

-0.36

POSITIVE LOGITS

 nahilalakip

0.70

ArrowToggle

0.64

 يتيمه

0.58

 HasFactory

0.56

AddTagHelper

0.52

RefNanny

0.49

تقاوى

0.49

 ExecuteAsync

0.48

XmlAccessorType

0.47

 ſind

0.47

Activations Density 0.002%

references to personal experiences and identities

my name

When text is a chat/forum/IRC speaker label (a username or nick often shown in angle brackets or as a line author), this neuron strongly activates.

No Comments

No Known Activations