INDEX

Explanations

numerical values and their associated contexts

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

code, math, and numbers

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

一

-0.44

intéress

-0.39

—

-0.39

Zuh

-0.37

 Suárez

-0.36

相

-0.35

 procura

-0.35

 intellectual

-0.35

zeß

-0.34

<sup>

-0.34

POSITIVE LOGITS

GEBURTSDATUM

0.85

 Administrativna

0.77

 autorytatywna

0.72

хьтан

0.71

Autoritní

0.69

utilisons

0.65

 otomatig

0.65

ioutil

0.64

Portale

0.61

RTLR

0.61

Activations Density 0.000%

numerical values and their associated contexts

code, math, and numbers

No Comments

No Known Activations