INDEX

Explanations

phrases that indicate reasons or justifications for events or actions

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

reasons for

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

 concor

-0.47

 peup

-0.46

 Dyer

-0.44

DetailView

-0.43

 menac

-0.43

 Ogden

-0.42

 гори

-0.42

gbar

-0.41

 Majefty

-0.41

Geno

-0.41

POSITIVE LOGITS

 varför

0.77

why

0.71

 hvorfor

0.66

 pourquoi

0.65

why

0.60

 warum

0.59

 waarom

0.59

 Pourquoi

0.58

Varför

0.57

Waarom

0.56

Activations Density 0.034%

phrases that indicate reasons or justifications for events or actions

reasons for

No Comments

No Known Activations