INDEX

Explanations

words or phrases that indicate legal authority or decisions

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

legal and factual information

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

 AspNetCore

-0.70

 Мексичка

-0.67

<unused41>

-0.65

<unused8>

-0.65

<unused47>

-0.65

<unused52>

-0.65

<unused74>

-0.65

 パンチラ

-0.65

[@BOS@]

-0.64

<unused3>

-0.64

POSITIVE LOGITS

CloseOperation

0.33

 Envío

0.32

 légale

0.31

 grabado

0.30

 Consultez

0.29

 leão

0.28

 sinta

0.28

 religieuse

0.27

 navidad

0.27

NSNumber

0.27

Activations Density 0.057%

words or phrases that indicate legal authority or decisions

legal and factual information

No Comments

No Known Activations