INDEX

Explanations

references to investigations or inquiries into actions and events related to the law

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 36-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_36/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.36.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

 estratég

-0.38

windowFixed

-0.35

etCode

-0.34

Spotted

-0.34

 auguri

-0.34

agoza

-0.34

 useContext

-0.34

publicain

-0.33

Chham

-0.33

 cameo

-0.33

POSITIVE LOGITS

 investigation

2.64

 investigations

2.44

investigation

2.30

 Investigation

2.27

Investigation

2.25

 investigate

2.25

investig

2.23

 investigating

2.22

 INVESTIGATION

2.14

 Investigations

2.13

Activations Density 0.685%

references to investigations or inquiries into actions and events related to the law

No Comments

No Known Activations