INDEX

Explanations

references to hierarchical structures or relationships, particularly involving "parent" objects or elements

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

region identifiers

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

 MainAxisSize

-0.39

 unre

-0.38

TagMode

-0.38

 <<<<<<<<<<<<<<

-0.37

 Winaray

-0.35

une

-0.30

 مشين

-0.29

 îna

-0.29

awaiter

-0.28

 much

-0.28

POSITIVE LOGITS

########.

0.59

 nakalista

0.57

Datuak

0.54

 Supervis

0.54

 harvester

0.53

SpringBootTest

0.52

formatics

0.52

abetes

0.52

 campsite

0.52

 kasarigan

0.51

Activations Density 0.020%

references to hierarchical structures or relationships, particularly involving "parent" objects or elements

region identifiers

No Comments

No Known Activations