INDEX

Explanations

punctuation and formatting characters

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

image references and descriptive labels

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

tagHelperRunner

-0.79

 мәкал

-0.75

EndContext

-0.69

 surla

-0.66

IndentedString

-0.66

 Italijani

-0.63

WriteTagHelper

-0.60

 gyhoeddwyd

-0.59

OGND

-0.59

 Chwiliwch

-0.57

POSITIVE LOGITS

 officielles

0.46

 visuales

0.46

 salvajes

0.44

 africains

0.44

 Figuren

0.42

 réunis

0.42

 sauvages

0.42

Principales

0.42

 fournis

0.41

 nationaux

0.41

Activations Density 0.021%

punctuation and formatting characters

image references and descriptive labels

No Comments

No Known Activations