INDEX

Explanations

references to sequences or positions in time

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

position in sequence

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

SequentialGroup

-0.45

ajur

-0.45

IsContent

-0.43

Hozzáférés

-0.41

 الرا

-0.40

 terbuka

-0.39

AccessFile

-0.38

Datuak

-0.38

 humanidad

-0.38

 betrek

-0.37

POSITIVE LOGITS

<?

0.46

новништво

0.45

Географиясе

0.45

Життєпис

0.43

❋

0.42

AlterField

0.42

⤹

0.42

 GenerationType

0.42

localVar

0.41

دانشنامهٔ

0.41

Activations Density 0.066%

references to sequences or positions in time

position in sequence

No Comments

No Known Activations