INDEX

Explanations

phrases indicating the presence of important concepts or impacts in a document

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 34-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_34/width_16k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.34.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 chose

-0.50

rungsseite

-0.47

 expressed

-0.45

 discovered

-0.44

 chofe

-0.43

EndContext

-0.43

 diagnosed

-0.43

 observed

-0.43

chose

-0.43

 wrote

-0.42

POSITIVE LOGITS

 been

0.91

Personensuche

0.65

 become

0.57

been

0.57

 relevance

0.56

 būti

0.54

 betrekking

0.54

levance

0.53

 být

0.52

 significance

0.51

Activations Density 0.359%

phrases indicating the presence of important concepts or impacts in a document

No Comments

No Known Activations