INDEX

Explanations

terms and phrases related to procedural and instructional contexts

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

illustrations, laser, relate, Decode

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

 airbags

-0.32

Toc

-0.31

 synag

-0.30

PRL

-0.28

 parce

-0.28

 tetrach

-0.27

了一种

-0.26

 taxis

-0.26

 heaters

-0.26

 umbi

-0.26

POSITIVE LOGITS

 Савезне

0.67



0.66

 <>",

0.64

 autorytatywna

0.63

errHandler

0.63

 мѣ

0.60

 Italijanski

0.60

BufferException

0.60

 فريبيس

0.60

 تضيفلها

0.60

Activations Density 0.003%

terms and phrases related to procedural and instructional contexts

illustrations, laser, relate, Decode

No Comments

No Known Activations