INDEX

Explanations

phrases indicating varying degrees of satisfaction or evaluation

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

Creating second

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

 prepare

-0.33

Tr

-0.33

 paik

-0.31

uxxxx

-0.31

prepare

-0.31

 Prepare

-0.31

 prepar

-0.30

 dommages

-0.30

สร

-0.30

chart

-0.29

POSITIVE LOGITS

ValueStyle

0.76

 صوتيه

0.67

 للمعارف

0.64

 للاسماء

0.63

 محفوظة

0.63

 Photocase

0.59

HasForeignKey

0.57

kuuta

0.55

MessageOf

0.55

 JAXBElement

0.53

Activations Density 0.002%

phrases indicating varying degrees of satisfaction or evaluation

Creating second

No Comments

No Known Activations