INDEX

Explanations

terms related to file and folder structures

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

folder

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

 Prest

-0.50

Ne

-0.47

Ay

-0.47

NA

-0.46

 EconPapers

-0.45

-0.44

 White

-0.44

Bem

-0.42

Ste

-0.42

 strike

-0.42

POSITIVE LOGITS

 Folder

1.24

 folder

1.16

 Folders

1.15

 folders

1.13

FOLDER

1.09

Folder

1.08

folder

1.05

folders

0.96

Folders

0.94

 Ordner

0.92

Activations Density 0.005%

terms related to file and folder structures

folder

No Comments

No Known Activations