INDEX

Explanations

technical terms related to neural network feature maps and performance metrics

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

feature map detection

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

 ContentValues

-0.48

tagHelperRunner

-0.44

Datuak

-0.43

usercontent

-0.43

orianCalendar

-0.40

contentLoaded

-0.39

turnstile

-0.39

Personendaten

-0.39

ьаж

-0.39

NewRow

-0.38

POSITIVE LOGITS

 architecture

0.48

textra

0.48

 architectures

0.46

 Architecture

0.44

ętr

0.44

SequentialGroup

0.43

 Bird

0.43

itecture

0.42

Bird

0.41

falla

0.40

Activations Density 0.027%

technical terms related to neural network feature maps and performance metrics

feature map detection

No Comments

No Known Activations