INDEX

Explanations

phrases related to instructions or procedural actions

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

pecabe

-0.56

IsPostBack

-0.55

 للاسماء

-0.52

 mijne

-0.50

 canadien

-0.50

########.

-0.49

 myſelf

-0.49

 śmier

-0.49

pingente

-0.49

httphttps

-0.49

POSITIVE LOGITS

InputTagHelper

0.40

 utafitiHapana

0.36

 InputDecoration

0.35

 conval

0.34

迎

0.33

My

0.32

XDECREF

0.31

kom

0.31

üli

0.31

twimg

0.29

Activations Density 0.178%

phrases related to instructions or procedural actions

No Comments

No Known Activations