INDEX

Explanations

terms related to academic presentations and publications

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

conference presentations and posters

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

 beſ

-0.36

 campañas

-0.32

 campanha

-0.32

 sağlar

-0.30

 básicas

-0.30

 deberían

-0.30

 anſ

-0.29

 zákon

-0.29

Provide

-0.29

 Medidas

-0.28

POSITIVE LOGITS



0.81

期刊论文

0.80

 presentations

0.72

 Presentations

0.68

 Administrativna

0.68

 estekak

0.68

 snippetHide

0.67

 presentation

0.64

contentLoaded

0.63

InputTagHelper

0.62

Activations Density 0.011%

terms related to academic presentations and publications

conference presentations and posters

No Comments

No Known Activations