INDEX

Explanations

terms related to CNN architecture and feature extraction

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

VGG (Very Deep Convolutional Networks)

np_acts-logits-general · gemini-2.0-flash

VGG networks

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_9/width_16k/average_l0_88

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.9.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

aduras

-0.31

 görün

-0.29

ALLAS

-0.28

 chores

-0.28

żesz

-0.27

 vectorielles

-0.27

țiunea

-0.27

 pharynx

-0.26

 weirdly

-0.26

ladung

-0.26

POSITIVE LOGITS

featureID

0.75

 propOrder

0.66

 surla

0.65

WithIOException

0.64

 EconPapers

0.63

parsedMessage

0.63

AnimationsModule

0.63

FormTagHelper

0.59

CppMethod

0.56

 configureStore

0.54

Activations Density 0.380%

terms related to CNN architecture and feature extraction

VGG (Very Deep Convolutional Networks)

VGG networks

No Comments

No Known Activations