INDEX

Explanations

words or phrases related to user actions or functionality in programming contexts

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

phrases related to using or implementing tools and processes in programming

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

Japanese and German place names

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

RIN

-0.55

erex

-0.53

volmente

-0.52

liga

-0.51

sc

-0.49

iglio

-0.48

thető

-0.48

yte

-0.48

lesse

-0.48

ését

-0.47

POSITIVE LOGITS

脚注の使い方

0.71

 المعيارى

0.69

DIPSETTING

0.68

 AssemblyVersion

0.66

 سكانية

0.63

المناصب

0.62

MENAFN

0.61

IsMutable

0.61

 JpaRepository

0.61

 ***!

0.60

Activations Density 0.017%

words or phrases related to user actions or functionality in programming contexts

phrases related to using or implementing tools and processes in programming

Japanese and German place names

No Comments

No Known Activations