INDEX

Explanations

terms related to complex concepts or phenomena that involve systems or structures

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

coined terms and phenomena

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

 baum

-0.45

 INTERESAR

-0.45

 GenerationType

-0.45

 linkovi

-0.44

دانشنامهٔ

-0.43

 CURIAM

-0.42

Chham

-0.41

UnusedPrivate

-0.41

RTLI

-0.40

albero

-0.40

POSITIVE LOGITS

 bekende

0.48

 phénomène

0.46

 coined

0.44

 oluş

0.44

VersionUID

0.43

 phenomenon

0.42

 fenomeno

0.40

 noqa

0.40

 Offisielt

0.39

 zprávy

0.39

Activations Density 0.056%

terms related to complex concepts or phenomena that involve systems or structures

coined terms and phenomena

No Comments

No Known Activations