INDEX

Explanations

phrases indicating past or ongoing research efforts and findings

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

"has" or "have" followed by a past participle

np_acts-logits-general · gemini-2.0-flash

past participle auxiliaries

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

 Knew

-0.50

 fays

-0.50

 osserva

-0.49

дём

-0.49

 pauv

-0.49

Was

-0.48

 femininos

-0.48

 carico

-0.48

knew

-0.48

 felic

-0.47

POSITIVE LOGITS

 been

1.86

 become

1.19

 gone

1.14

 gotten

1.11

been

1.02

had

1.01

 come

0.98

got

0.97

 never

0.94

 taken

0.94

Activations Density 0.441%

phrases indicating past or ongoing research efforts and findings

"has" or "have" followed by a past participle

past participle auxiliaries

No Comments

No Known Activations