INDEX

Explanations

specific identifiers, metrics, and terminology often related to academic and health-related contexts

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

 surla

-0.78

 queſta

-0.72

tanleria

-0.68

 ſei

-0.62

 geſch

-0.61

 ſind

-0.61

:✨

-0.60

SharedDtor

-0.60

iſchen

-0.59

 ſehen

-0.59

POSITIVE LOGITS

ValueStyle

0.40

يميديا

0.35

 Voulez

0.33

).

0.30

0.29

 Cliquez

0.28

])

0.28

})}\

0.28

 comigo

0.28

 selon

0.28

Activations Density 0.108%

specific identifiers, metrics, and terminology often related to academic and health-related contexts

No Comments

No Known Activations