INDEX

Explanations

terms related to performance, usage, or qualities of systems and products

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

存于互联网档案馆

-0.52

TypedDataSet

-0.51

tomans

-0.50

ﷺ

-0.48

وردار

-0.45

impresa

-0.44

stasy

-0.44

voorbeeld

-0.44

리가

-0.43

jenigen

-0.43

POSITIVE LOGITS

 houſe

0.59

 Houſe

0.55

 quæ

0.49

0.46

 auffi

0.46

 Monfieur

0.45

 déc

0.45

ſelves

0.44

 faſt

0.44

<?,

0.42

Activations Density 0.154%

terms related to performance, usage, or qualities of systems and products

No Comments

No Known Activations