INDEX

Explanations

expressions related to considerations or opinions

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

references to perceptions or opinions about events or issues

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

in regard to or regarded as

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-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

Искәрмәләр

-0.77

 <>",

-0.77

TokenNameLBRACE

-0.74

laughs

-0.69

vábbi

-0.67

oretical

-0.67

ContentLoaded

-0.66

cestors

-0.65

EndProject

-0.65

WebServlet

-0.64

POSITIVE LOGITS

ing

0.71

 مرئيه

0.65

rous

0.59

thalam

0.57

<th>

0.55

}),

0.55

"__

0.52

 REGARD

0.52

*/].

0.51

ும்

0.51

Activations Density 0.025%

expressions related to considerations or opinions

references to perceptions or opinions about events or issues

in regard to or regarded as

No Comments

No Known Activations