INDEX

Explanations

expressions conveying perceptions or evaluations

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

phrases indicating evaluation or condition

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

sounds like or looks like

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

 كومونز

-0.52

 endings

-0.48

헌

-0.47

Bissau

-0.46

 tarihinde

-0.45

Чыганаклар

-0.44

 włosów

-0.44

 throughout

-0.43

addLast

-0.43

SPECIFIED

-0.43

POSITIVE LOGITS

 متعلقه

0.76

 nahilalakip

0.74

पया

0.74

 BoxFit

0.72

getMenuInflater

0.69

 InputDecoration

0.68

styleType

0.67

 Walkover

0.67

 interesting

0.67

Interesting

0.65

Activations Density 0.029%

expressions conveying perceptions or evaluations

phrases indicating evaluation or condition

sounds like or looks like

No Comments

No Known Activations