INDEX

Explanations

words indicating frequency and intensity

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

predominant, too much, actively pursuing

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

 ſind

-0.54

TagHelper

-0.50

 fashiola

-0.50

UnusedPrivate

-0.49

 GenerationType

-0.49

 célè

-0.48

enablog

-0.48

<unused8>

-0.48

[@BOS@]

-0.48

<unused14>

-0.48

POSITIVE LOGITS

principalColumn

0.44

 Италијани

0.42

WithIOException

0.39

']==

0.39

省市镇

0.38

indd

0.36

 дописавши

0.36

uxxxx

0.36

Ӕ

0.35

:✨

0.35

Activations Density 0.038%

words indicating frequency and intensity

predominant, too much, actively pursuing

No Comments

No Known Activations