INDEX

Explanations

non-textual characters or formatting in the document

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

Code snippets or non-English characters

np_acts-logits-general · gemini-2.0-flash

copyright notice

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

хьтан

-0.64

 informée

-0.56

 ProtoMessage

-0.53

intios

-0.53

脚注の使い方

-0.51

Espèce

-0.50

tanleria

-0.50

ombes

-0.50

RenderAtEndOf

-0.49

UAGES

-0.48

POSITIVE LOGITS

0.89

0.79

0.70

0.64

 Schwer

0.47

 NATURE

0.42

0.41

 Brink

0.41

 VIRTUAL

0.41

 dulce

0.40

Activations Density 0.283%

non-textual characters or formatting in the document

Code snippets or non-English characters

copyright notice

No Comments

No Known Activations