INDEX

Explanations

terms related to safety, health, and support mechanisms

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-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

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

 ModelExpression

-0.62

complexContent

-0.58

 useAppContext

-0.57

tanleria

-0.56

RegistryLite

-0.53

 encarga

-0.52

 oprot

-0.52

 meille

-0.50

şört

-0.50

 AssemblyCompany

-0.50

POSITIVE LOGITS

httphttps

0.35

 Some

0.35

 devices

0.35

 some

0.34

Pu

0.34

Tikang

0.33

some

0.33

 SOME

0.32

Gul

0.32

yes

0.32

Activations Density 0.071%

terms related to safety, health, and support mechanisms

No Comments

No Known Activations