INDEX

Explanations

punctuation and formatting used in technical documentation

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

safety and toxicity studies

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

 saites

-0.56

󠁴

-0.48

 ujednoznacz

-0.42

 Connexion

-0.42

DropTable

-0.42

loroethene

-0.40

 Roskov

-0.40

SpringRunner

-0.39

ölkerung

-0.39

 Metcalf

-0.38

POSITIVE LOGITS

 <<<<<<<<<<<<<<

0.58

 disponibilités

0.56

ImageContext

0.49

DataAnnotations

0.47

:✨

0.47

httphttps

0.47

ValueStyle

0.45

 gyhoeddwyd

0.41

ARROLL

0.41

RTSC

0.41

Activations Density 0.001%

punctuation and formatting used in technical documentation

safety and toxicity studies

No Comments

No Known Activations