INDEX

Explanations

elements related to legal and federal terms or procedures

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

conviction

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

 protoimpl

-0.53

 ModelExpression

-0.51

 known

-0.50

مصادر

-0.50

known

-0.44

アンス

-0.44

coefficient

-0.43

웻

-0.43

dür

-0.41

.*")]

-0.41

POSITIVE LOGITS

 jsPsych

0.57

httphttps

0.55

Aholisi

0.46

Portale

0.46

UserScript

0.45

ContentAsync

0.43

 GenerationType

0.40

zzlies

0.40

Hauptartikel

0.39

GenerationType

0.38

Activations Density 0.149%

elements related to legal and federal terms or procedures

conviction

No Comments

No Known Activations