INDEX

Explanations

conjunctions and their usage in connecting ideas or phrases

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

technical terms, programming

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

 kasarigan

-0.52

 pinulongan

-0.48

 équilibr

-0.45

 vectorielles

-0.44

 Einfluß

-0.42

 działal

-0.42

 Anhäng

-0.42

 dieses

-0.41

 sedangkan

-0.40

 działalności

-0.40

POSITIVE LOGITS

PMailer

0.59

protoimpl

0.49

 protoimpl

0.44

CPtr

0.44

SequentialGroup

0.44

 rhestr

0.43

WebServlet

0.42

蚪

0.41

nodoc

0.41

thâu

0.41

Activations Density 0.074%

conjunctions and their usage in connecting ideas or phrases

technical terms, programming

No Comments

No Known Activations