INDEX

Explanations

references to physical or logical locations

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

Attribute name followed by equals sign

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

="<?=

-0.50

endpush

-0.49

endforeach

-0.48

 Stuttgart

-0.46

{?>

-0.46

<eos>

-0.45

endish

-0.44

 BaseEntity

-0.44

 biex

-0.44

 ['./

-0.43

POSITIVE LOGITS

findpost

0.86

Location

0.66

DockStyle

0.65

 AssemblyTitle

0.63

setVerticalGroup

0.63

complexContent

0.61

 Location

0.60

 propOrder

0.60

 defaultstate

0.59

цездатний

0.57

Activations Density 0.000%

references to physical or logical locations

Attribute name followed by equals sign

No Comments

No Known Activations