INDEX

Explanations

references to types of panels and networks in scientific contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

róleo

-0.53

 frontières

-0.52

graphiques

-0.51

 négociations

-0.51

Vertical

-0.51

 Italijanski

-0.49

salvar

-0.49

 Naciones

-0.49

лтемелер

-0.49

calça

-0.48

POSITIVE LOGITS

of

0.56

 handful

0.43

 урна

0.42

addPreferredGap

0.41

 bunch

0.41

undred

0.41

unately

0.41

 dozens

0.40

ftagPool

0.40

fjspx

0.40

Activations Density 0.099%

references to types of panels and networks in scientific contexts

No Comments

No Known Activations