INDEX

Explanations

phrases related to communication, law, and eyewitness accounts

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

as

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

 angew

-0.40

 Polsek

-0.40

werf

-0.39

 Krakowie

-0.36

 Polisi

-0.36

 Varma

-0.35

 Napoli

-0.35

 empe

-0.35

 Andromeda

-0.35

 opge

-0.34

POSITIVE LOGITS

 Wikimedijinoj

0.54

تقاوى

0.49

MethodManager

0.45

SBATCH

0.45

Jegyzetek

0.45

\{\\

0.43

 ब्रेकडाउन

0.43

readFileSync

0.43

SPATH

0.42

 Италијани

0.42

Activations Density 0.142%

phrases related to communication, law, and eyewitness accounts

as

No Comments

No Known Activations