INDEX

Explanations

terms related to plain formats and circuits

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_0/width_65k/average_l0_73

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.0.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

 Cæsar

-0.90

 Jérusalem

-0.85

 Jurí

-0.81

 navideño

-0.76

 itſelf

-0.75

 Cleopatra

-0.74

 Shakspeare

-0.74

TagMode

-0.72

 warszawa

-0.71

脚注の使い方

-0.71

POSITIVE LOGITS

 plain

0.90

Cir

0.79

plain

0.75

Plain

0.69

cir

0.66

 Plain

0.63

cir

0.62

 onCreateView

0.60

Lam

0.56

Cir

0.55

Activations Density 0.127%

terms related to plain formats and circuits

No Comments

No Known Activations