INDEX

Explanations

references to possession and relationships in various contexts

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

our results / program mode

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

te

-0.29

 מלא

-0.26

 thereof

-0.26

yczą

-0.25

 regardé

-0.24

 dirait

-0.24

 vorge

-0.24

 bộ

-0.23

分で

-0.23

POSITIVE LOGITS

OGND

0.91

parsedMessage

0.90

 nakalista

0.81

 myſelf

0.74

ロウィン

0.73

PerformLayout

0.73

featureID

0.72

expandindo

0.71

 étoient

0.71

 ſelbſt

0.70

Activations Density 0.023%

references to possession and relationships in various contexts

our results / program mode

No Comments

No Known Activations