INDEX

Explanations

phrases indicating potential actions or accomplishments

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

 rangs

-0.39

vejec

-0.35

 tends

-0.34

oblotting

-0.34

 tend

-0.33

tend

-0.33

 seniority

-0.33

 viewWillAppear

-0.33

 bottes

-0.32

 routines

-0.32

POSITIVE LOGITS

CanBe

0.79

could

0.70

Could

0.69

 Could

0.68

LookAnd

0.67

Bisa

0.66

Possibly

0.66

potentially

0.66

Puedes

0.65

 Podemos

0.65

Activations Density 0.010%

phrases indicating potential actions or accomplishments

No Comments

No Known Activations