INDEX

Explanations

specific nouns or identifiers related to technology and users

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

citations with symbols like @

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

 EconPapers

-0.71

IntoConstraints

-0.68

 Económica

-0.63

 continúas

-0.62

 المعيارى

-0.61

 Photocase

-0.60

 vermelhas

-0.60

Autoritní

-0.58

parsedMessage

-0.58

 Normdatei

-0.57

POSITIVE LOGITS

 hashtag

0.47

0.42

#@

0.36

ApiModelProperty

0.33

/#

0.33

alg

0.32

anglès

0.31

//#

0.31

Activations Density 0.229%

specific nouns or identifiers related to technology and users

citations with symbols like @

No Comments

No Known Activations