INDEX

Explanations

phrases related to requests for user input or feedback

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

repetition or doing something again

np_acts-logits-general · gemini-2.0-flash

re-start, repeat, again

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

ngdoc

-0.54

 HttpNotFound

-0.52

hearsed

-0.52

 lateinit

-0.51

цездатний

-0.50

└──

-0.50

spyOn

-0.47

velope

-0.46

enedor

-0.46

 répondit

-0.46

POSITIVE LOGITS

 again

1.54

 yeniden

1.54

重新

1.52

 opnieuw

1.50

 ulang

1.50

 repeat

1.50

 다시

1.49

re

1.49

 tekrar

1.46

 повтор

1.44

Activations Density 0.844%

phrases related to requests for user input or feedback

repetition or doing something again

re-start, repeat, again

No Comments

No Known Activations

phrases related to requests for user input or feedback

repetition or doing something again

re-start, repeat, again

No Comments

No Known Activations