INDEX

Explanations

expressions and questions regarding reasons and guidelines

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

why and why-to

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

 purpoſe

-0.63

 uſe

-0.62

ſehen

-0.62

 uſed

-0.61

 محفوظة

-0.60

 dieſe

-0.60

رشف

-0.59

 leaſt

-0.59

ſelves

-0.59

 ſeine

-0.59

POSITIVE LOGITS

toBeTruthy

0.45

 intricacies

0.39

done

0.39

readystatechange

0.35

 done

0.34

 Redacción

0.33

 hushed

0.33

Done

0.33

 poser

0.32

Redacción

0.32

Activations Density 0.023%

expressions and questions regarding reasons and guidelines

why and why-to

No Comments

No Known Activations