INDEX

Explanations

the word "à" in various contexts

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

 honneur

-0.54

 Púb

-0.51

 nôtre

-0.49

 jugement

-0.48

 lequel

-0.47

 lisäksi

-0.47

Slf

-0.46

 todėl

-0.44

 exécu

-0.44

 rangement

-0.44

POSITIVE LOGITS

0.80

à

0.62

 goTo

0.61

AnchorTagHelper

0.60

at

0.59

Σε

0.57

 μην

0.57

to

0.56

aat

0.56

goTo

0.56

Activations Density 0.012%

the word "à" in various contexts

No Comments

No Known Activations