INDEX

Explanations

discussions about past events, particularly related to intelligence and geopolitical issues

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

words after speech or names

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

 Autorisations

-0.48

 élas

-0.47

 kran

-0.45

 BoxFit

-0.45

期刊论文

-0.45

(!__

-0.44

OGND

-0.43

Tikang

-0.43

Tracce

-0.42

 Plastik

-0.42

POSITIVE LOGITS

 kanssa

0.42

 становника

0.42

föra

0.35

ptonshire

0.35

 mondta

0.34

ongiorno

0.34

 sguardo

0.33

før

0.33

 hinweg

0.33

warten

0.33

Activations Density 0.056%

discussions about past events, particularly related to intelligence and geopolitical issues

words after speech or names

No Comments

No Known Activations