INDEX

Explanations

specific numerical and geographical data points within a context

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

federal and state divisions

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

qiao

-0.35

antenna

-0.35

 coklat

-0.32

浴び

-0.32

ngths

-0.31

外

-0.30

IDENTIFIED

-0.30

 infeksi

-0.30

 puntas

-0.30

klärung

-0.30

POSITIVE LOGITS

 interstate

0.60

 federal

0.60

 Federal

0.58

 queſta

0.58

 myſelf

0.58

 nationwide

0.58

 ſou

0.55

Federal

0.54

ſelf

0.54

 ſta

0.54

Activations Density 0.079%

specific numerical and geographical data points within a context

federal and state divisions

No Comments

No Known Activations