INDEX

Explanations

terms related to safety and protection measures

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 khăn

-0.39

 getattr

-0.39

 chair

-0.39

 studio

-0.37

karan

-0.36

 litro

-0.35

 Ganzen

-0.35

 Malabar

-0.35

 trebui

-0.35

 chorus

-0.35

POSITIVE LOGITS

AddTagHelper

0.53

 typelib

0.52

 bezpiecze

0.51

 safety

0.50

 Roskov

0.48

ValueStyle

0.46

 keselamatan

0.46

 transfieras

0.46

PMailer

0.45

 protect

0.45

Activations Density 0.165%

terms related to safety and protection measures

No Comments

No Known Activations