INDEX

Explanations

phrases that express authority and knowledge, particularly those relating to power dynamics and information access

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

octanol

-0.37

 réception

-0.37

 wikipagina

-0.37

 Clam

-0.35

 hodnot

-0.35

 konusu

-0.35

 verein

-0.35

 SwitchCompat

-0.35

zelfde

-0.34

erees

-0.34

POSITIVE LOGITS

mergeFrom

0.55

 possessed

0.50

HasForeignKey

0.48

poss

0.48

 possession

0.47

 expertise

0.47

Poss

0.46

 препратки

0.45

">//

0.45

 Poss

0.45

Activations Density 0.114%

phrases that express authority and knowledge, particularly those relating to power dynamics and information access

No Comments

No Known Activations