INDEX

Explanations

occurrences of the number zero across various contexts

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

numbers and magnitudes

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

 assist

-0.59

 iscri

-0.55

 assisting

-0.53

itism

-0.48

bibfnamefont

-0.46

 ASSIST

-0.44

 assists

-0.44

 blij

-0.43

 personagem

-0.42

حياتها

-0.42

POSITIVE LOGITS

0.90

Thousand

0.65

 XNUMX

0.64

 Millennium

0.64

qian

0.60

 bookId

0.58

thousand

0.58

 Thousand

0.57

 thousand

0.57

０

0.57

Activations Density 0.019%

occurrences of the number zero across various contexts

numbers and magnitudes

No Comments

No Known Activations