INDEX

Explanations

phrases and concepts related to algorithms and technical discussions

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

AFAIK or if your

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

 للمعارف

-0.80

 ſind

-0.73

 queſta

-0.72

 defaultstate

-0.70

Datuak

-0.70

bootstrapcdn

-0.69

 zwiſchen

-0.68

 témoig

-0.67

pecabe

-0.67

 indígen

-0.67

POSITIVE LOGITS

ValueStyle

0.33

 Niedersachsen

0.33

igshid

0.30

0.29

ar

0.28

<eos>

0.28

 Maher

0.28

Serializer

0.27

 新增

0.27

redux

0.27

Activations Density 0.019%

phrases and concepts related to algorithms and technical discussions

AFAIK or if your

No Comments

No Known Activations