INDEX

Explanations

statistics and comparisons between metrics or measurements

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

modern methods, men

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

enumii

-0.47

 estekak

-0.43

InjectAttribute

-0.43

IndentedString

-0.41

脚注の使い方

-0.40

lids

-0.40

ContainerState

-0.40

TextBoxColumn

-0.40

 Ecuador

-0.39

brug

-0.39

POSITIVE LOGITS

ValueStyle

0.54

новништво

0.52

frontmatter

0.42

tanleria

0.38

+#+

0.37

AndEndTag

0.36

 przecież

0.35

ArgsConstructor

0.35

 متعلقه

0.35

pexpr

0.34

Activations Density 0.080%

statistics and comparisons between metrics or measurements

modern methods, men

No Comments

No Known Activations