INDEX

Explanations

references to returning or going back to previous topics or texts

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

return to text

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

ין

-0.35

ujących

-0.27

으며

-0.26

Quic

-0.26

 successor

-0.25

 Eintritt

-0.25

 semacam

-0.25

dean

-0.25

 sûr

-0.25

 sikkert

-0.24

POSITIVE LOGITS

 المعيارى

0.94

+#+#

0.79

 ویکی‌آمباردا

0.78

 StringTokenizer

0.75

脚注の使い方

0.73

 MainAxisSize

0.70

ロウィン

0.67

consulté

0.66

sizeCache

0.65

<unused16>

0.63

Activations Density 0.027%

references to returning or going back to previous topics or texts

return to text

No Comments

No Known Activations