INDEX

Explanations

instances of comparative language and quantifications in descriptions

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

Followed by end-of-turn or start-of-turn tokens

np_acts-logits-general · gemini-2.0-flash

exhaustive explicit small scale computations

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

ंदीखरीदारी

-0.39

 durs

-0.36

xase

-0.35

ref

-0.35

XMLSchema

-0.34

spyOn

-0.34

 form

-0.34

 corporation

-0.33

SourceChecksum

-0.33

IV

-0.33

POSITIVE LOGITS

الحياه

0.77

 فريبيس

0.76

 BoxFit

0.63

 Taktlose

0.61

 للمعارف

0.60

 Bewußt

0.59

 Administrativna

0.59

 وتسجيلات

0.57

 Roskov

0.56

tagHelperRunner

0.55

Activations Density 0.460%

instances of comparative language and quantifications in descriptions

Followed by end-of-turn or start-of-turn tokens

exhaustive explicit small scale computations

No Comments

No Known Activations