INDEX

Explanations

references to discussions and comments related to research or reviews

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

respective collaborators

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

󠁢

-0.65

-0.62

isNaN

-0.49

astéroïdes

-0.47

ronpa

-0.45

 BoxFit

-0.44

moral

-0.43

iffen

-0.43

-------------</

-0.43

Clik

-0.42

POSITIVE LOGITS

 respectively

0.63

 respectivamente

0.57

 respectivement

0.55

各有

0.54

 reputations

0.51

 themselves

0.49

 respective

0.48

 versions

0.48

それぞれ

0.47

respectively

0.46

Activations Density 0.135%

references to discussions and comments related to research or reviews

respective collaborators

No Comments

No Known Activations