INDEX

Explanations

references to specific things or ideas in discourse

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

 êtres

-0.48

SpringBootTest

-0.43

 voorbere

-0.42

endphp

-0.41

 Vorbereitung

-0.40

ContentLoaded

-0.40

 zicht

-0.39

 sfeer

-0.38

 głó

-0.37

 goederen

-0.37

POSITIVE LOGITS

 nasty

0.57

 stuff

0.57

 horrible

0.55

 kaarangay

0.52

 horrid

0.51

 bogus

0.51

 '\\;'

0.50

あんな

0.50

box

0.49

 junk

0.49

Activations Density 0.050%

references to specific things or ideas in discourse

No Comments

No Known Activations