INDEX

Explanations

numerical values associated with specific cases or fields in a structured data format

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

field numbers or indices

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

ConstraintMaker

-0.79

IUrlHelper

-0.79

uxxxx

-0.73

EndProject

-0.71

 detailsNormal

-0.71

 يتيمه

-0.71

DoubleQuotes

-0.70

endphp

-0.68

addGap

-0.66

ſehen

-0.66

POSITIVE LOGITS

 Utama

0.32

 Allgemeine

0.31

 primul

0.31

farande

0.29

的第一

0.29

 первой

0.29

最初に

0.29

 pierwsze

0.29

 angefangen

0.28

 pierwszym

0.28

Activations Density 0.136%

numerical values associated with specific cases or fields in a structured data format

field numbers or indices

No Comments

No Known Activations