INDEX

Explanations

references to individuals and their roles or actions within a corporate or legal context

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

names and titles

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

 beroep

-0.42

 randomized

-0.40

 commerciaux

-0.40

 koszulka

-0.39

 jurk

-0.39

 czeka

-0.39

 organisatie

-0.38

 bisnis

-0.38

 extranjera

-0.38

 turística

-0.38

POSITIVE LOGITS

Personendaten

0.56

 الحره

0.53

featureID

0.51

 surla

0.50

 GenerationType

0.50

IUrlHelper

0.49

 kasarigan

0.48

 nakalista

0.47

LLocation

0.47

MLLoader

0.46

Activations Density 0.091%

references to individuals and their roles or actions within a corporate or legal context

names and titles

No Comments

No Known Activations