INDEX

Explanations

occurrences of the word "Bo" in various contexts, suggesting a focus on identifying the character or element related to that word

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

 Geduld

-0.58

 Cataluña

-0.54

graphique

-0.53

 mistrzost

-0.52

 catedral

-0.52

 umani

-0.51

reference

-0.49

 mieszkań

-0.48

 warszaw

-0.48

 εμφ

-0.48

POSITIVE LOGITS

Bo

1.20

Bo

1.14

BO

1.06

bo

1.05

bo

0.96

BO

0.96

Boe

0.83

 Boeh

0.79

 Boone

0.78

 Boiler

0.73

Activations Density 0.008%

occurrences of the word "Bo" in various contexts, suggesting a focus on identifying the character or element related to that word

No Comments

No Known Activations