INDEX

Explanations

phrases that indicate identity or nicknames

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 38-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_38/width_16k/average_l0_128

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.38.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

 conversión

-0.40

 Gemeins

-0.38

 amortigu

-0.38

 Bedürf

-0.35

 Höchst

-0.35

 iniciativa

-0.35

 conmemor

-0.35

 nordrhein

-0.35

 Einrichtungen

-0.35

 Verwaltungs

-0.34

POSITIVE LOGITS

 Monfieur

0.77

iſche

0.75

ftagPool

0.74

 Uncle

0.73

 パンチラ

0.73

>--}}

0.72

 vooz

0.72

 Bubba

0.71

<unused76>

0.71

<unused43>

0.71

Activations Density 0.279%

phrases that indicate identity or nicknames

No Comments

No Known Activations