INDEX

Explanations

specific nouns and numbers that indicate measurements or contexts

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

Canning, cells, farmers, subgroup Ib

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

skirts

-0.43

jacking

-0.41

 Audiodateien

-0.40

IANGLES

-0.40

🏾

-0.39

 utafitiHapana

-0.39

🇶

-0.38

 baum

-0.37

Lic

-0.37

teenth

-0.37

POSITIVE LOGITS

 MenuView

0.45

 recall

0.40

BeginInit

0.40

 Recall

0.39

recall

0.39

 للاسماء

0.38

ignano

0.38

EndGlobalSection

0.38

下载附件

0.37

 dés

0.37

Activations Density 0.160%

specific nouns and numbers that indicate measurements or contexts

Canning, cells, farmers, subgroup Ib

No Comments

No Known Activations