INDEX

Explanations

words related to changes or effects in research contexts

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

Preceding "able", "associated", or "important"

np_acts-logits-general · gemini-2.0-flash

was able / incapable

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 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

belong

-0.84

 belong

-0.78

 belonging

-0.70

 belonged

-0.70

 apparti

-0.69

Efq

-0.68

 belongs

-0.66

 Belong

-0.65

 appartenant

-0.65

 appartiennent

-0.63

POSITIVE LOGITS

 able

0.76

 مرئيه

0.60

jooq

0.56

 SqlParameter

0.55

+#+#

0.53

 capable

0.52

قایناق‌لار

0.50

èvement

0.50

!*\

0.49

شهاد

0.48

Activations Density 0.628%

words related to changes or effects in research contexts

Preceding "able", "associated", or "important"

was able / incapable

No Comments

No Known Activations