INDEX

Explanations

phrases that indicate intention or influence

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

should govern or influence

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

 OnInit

-0.46

UnusedPrivate

-0.42

 Theſe

-0.39

 AppDelegate

-0.39

SBATCH

-0.39

 auroit

-0.38

HandleFunc

-0.38

 gewissen

-0.36

ต่

-0.36

espère

-0.36

POSITIVE LOGITS

󠁣

0.49

UrlResolution

0.47

LLocation

0.44

@@@@@

0.44

хьтан

0.42

Captor

0.42

WebServlet

0.42

kuuta

0.41

hoeddwyd

0.40

CWE

0.40

Activations Density 0.128%

phrases that indicate intention or influence

should govern or influence

No Comments

No Known Activations