INDEX

Explanations

sequences that appear to be formatted or structured text, such as code comments or document sections

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

code comments and licenses

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

ContentLoaded

-0.80

fjspx

-0.74

XmlAccessType

-0.65

AnchorStyles

-0.65

]")]

-0.64

.*")]

-0.63

 MessageBoxIcon

-0.61

contentLoaded

-0.60

 &___

-0.57

DMETHOD

-0.56

POSITIVE LOGITS

httphttps

0.60

 faſt

0.37

 fubject

0.36

 ſta

0.34

 abſ

0.32

 resourceCulture

0.32

 Équipe

0.31

AddTagHelper

0.30

 ſtate

0.29

 entreprise

0.29

Activations Density 0.073%

sequences that appear to be formatted or structured text, such as code comments or document sections

code comments and licenses

No Comments

No Known Activations