INDEX

Explanations

specific patterns related to video links or identifiers

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

URLs and video links

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

 ſtate

-0.50

 ſche

-0.50

 juſ

-0.50

 prêtre

-0.48

 ſen

-0.47

 perſon

-0.47

 équi

-0.47

 ſou

-0.47

 ſta

-0.45

 ſtre

-0.45

POSITIVE LOGITS

 betweenstory

0.91

+#+

0.60

0.57

enumi

0.56

 YouTube

0.54

 AssemblyCulture

0.53

hyrchwyd

0.53

 slideshow

0.53

youtube

0.52

 youtube

0.52

Activations Density 0.001%

specific patterns related to video links or identifiers

URLs and video links

No Comments

No Known Activations