INDEX

Explanations

requests or calls to action prefaced by the word "please."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 AssemblyTitle

-0.49

 fieldNum

-0.49

berdayakan

-0.47

NCP

-0.46

HandleFunc

-0.45

 againſt

-0.45

GSC

-0.44

ICM

-0.44

MCM

-0.44

Portale

-0.43

POSITIVE LOGITS

 please

0.96

Please

0.96

please

0.94

 Please

0.94

 Bitte

0.81

PLEASE

0.79

 PLEASE

0.77

Bitte

0.73

 bitte

0.72

请

0.65

Activations Density 0.042%

requests or calls to action prefaced by the word "please."

No Comments

No Known Activations