INDEX

Explanations

specific terms related to locations, features, and actions associated with communities or settings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

IVEREF

-0.61

ⓧ

-0.60

PerformLayout

-0.53

 beſte

-0.52

amssymb

-0.52

 ſeine

-0.50

 ſei

-0.49

 müſſen

-0.47



-0.47

PreExecute

-0.46

POSITIVE LOGITS

 also

0.47

 inoltre

0.40

 moreover

0.39

 Мексичка

0.37

ClientSize

0.36

 też

0.36

Cyfarwyddwr

0.35

 juga

0.35

 tampoco

0.33

 propOrder

0.33

Activations Density 0.913%

specific terms related to locations, features, and actions associated with communities or settings

No Comments

No Known Activations