INDEX

Explanations

phrases that indicate references or attributions to sources

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

 estekak

-0.67

parsedMessage

-0.55

MultipartFile

-0.48

jenigen

-0.45

 Händen

-0.45

batore

-0.45

 Öffentlichkeit

-0.44

 język

-0.44

 Carbide

-0.43

 Vermögens

-0.42

POSITIVE LOGITS

 לפי

0.90

 وفق

0.80

 ตาม

0.78

 Secondo

0.76

Según

0.74

 theo

0.72

ตาม

0.71

Secondo

0.69

 according

0.69

 accordance

0.68

Activations Density 0.027%

phrases that indicate references or attributions to sources

No Comments

No Known Activations