INDEX

Explanations

phrases indicating a potential risk or threat

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-pt-res/layer_2/width_16k/average_l0_67

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.2.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

 poussière

-0.57

 mercad

-0.45

########.

-0.43

 seng

-0.43

quelize

-0.41

Baylor

-0.41

 unknownFields

-0.40

 fallu

-0.39

Manfaat

-0.39

TabStop

-0.38

POSITIVE LOGITS

 pose

1.33

 poses

1.13

 posed

1.12

 posing

1.10

 Pose

1.09

Pose

1.02

 threatening

0.86

 threaten

0.84

 threatens

0.81

 pretend

0.76

Activations Density 0.191%

phrases indicating a potential risk or threat

No Comments

No Known Activations