INDEX

Explanations

references to policies, rules, and instructions within a document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 otomatig

-0.70

tonode

-0.65

.*")]

-0.63

 protoimpl

-0.58

actionMode

-0.57

Tikang

-0.55

 saites

-0.54

GOTREF

-0.53

 فريبيس

-0.51

Билгалдахарш

-0.50

POSITIVE LOGITS

 therefore

0.52

 derfor

0.47

Therefore

0.44

 Therefore

0.44

 perciò

0.44

Поэтому

0.43

 joten

0.43

you

0.43

этому

0.42

 todėl

0.42

Activations Density 0.205%

references to policies, rules, and instructions within a document

No Comments

No Known Activations