INDEX

Explanations

numerical values and their relationships in a data-centric context

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

9hr sequences

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

__.__

-0.31

__()

-0.31

 Gegensatz

-0.30

 lehetős

-0.30

 greatly

-0.28

migrationBuilder

-0.27

RTCK

-0.27

 pagkak

-0.27

 względu

-0.26

 régal

-0.26

POSITIVE LOGITS

 betweenstory

0.70

 linkovi

0.60

 Verſ

0.60

[@BOS@]

0.59

<unused79>

0.59

<unused16>

0.58

<unused17>

0.58

<unused14>

0.58

<unused3>

0.58

<unused8>

0.58

Activations Density 0.006%

numerical values and their relationships in a data-centric context

9hr sequences

No Comments

No Known Activations