INDEX

Explanations

proper nouns and references from various contexts, including names and products

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

precum treat

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

AndEndTag

-0.71

 Administrativna

-0.68

 autorytatywna

-0.62

 EconPapers

-0.59

 snippetHide

-0.57

참고

-0.56

awaiter

-0.56

IntoConstraints

-0.56

IContainer

-0.56

SequentialGroup

-0.56

POSITIVE LOGITS

 betweenstory

0.53

fromnode

0.39

 consiguió

0.39

 traslad

0.38

 trasladado

0.38

 Rahman

0.38

istream

0.38

0.37

こんにちは

0.36

 hormigón

0.36

Activations Density 0.236%

proper nouns and references from various contexts, including names and products

precum treat

No Comments

No Known Activations