INDEX

Explanations

references to breaking news and urgent news updates

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

breaking news

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

évaluateur

-0.48

 externes

-0.39

AnchorTagHelper

-0.38

 sostenibilidad

-0.38

 terapé

-0.37

nictwa

-0.36

Methods

-0.36

 convivencia

-0.35

tanleria

-0.35

ensement

-0.35

POSITIVE LOGITS

breaking

0.85

BREAKING

0.74

Breaking

0.66

 Breaking

0.65

 breaking

0.65

 BREAKING

0.64

 news

0.62

emergency

0.58

PullParser

0.58

urgent

0.58

Activations Density 0.005%

references to breaking news and urgent news updates

breaking news

No Comments

No Known Activations