INDEX

Explanations

counts or quantities associated with time or intervals

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

durations in hours

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

writeFieldEnd

-0.49

 istrinya

-0.48

 Gästen

-0.46

 miniaturka

-0.46

 mesmos

-0.44

 GenerationType

-0.44

 suaminya

-0.44

 rumahnya

-0.44

 ninguém

-0.43

affaire

-0.43

POSITIVE LOGITS

onedDateTime

0.53

 months

0.50

 days

0.49

GEBURTSDATUM

0.48

 weeks

0.47

 strtotime

0.46

 hours

0.44

…]

0.43

 poffible

0.43

 transfieras

0.43

Activations Density 0.025%

counts or quantities associated with time or intervals

durations in hours

No Comments

No Known Activations