INDEX

Explanations

tokenized programming terms and code-related structures

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

Kyurangers or related terms

np_acts-logits-general · gemini-2.0-flash

character followed by action

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

always

-0.66

 همیشه

-0.65

Always

-0.63

 always

-0.61

 irré

-0.61

 Always

-0.61

 alltid

-0.61

ALWAYS

-0.61

ScopeManager

-0.60

 préférence

-0.60

POSITIVE LOGITS

 attack

0.71

 confront

0.70

 interrog

0.69

 chase

0.65

 attacked

0.65

 rejoin

0.64

 teleport

0.63

 hear

0.63

 interro

0.63

 encounter

0.63

Activations Density 0.251%

tokenized programming terms and code-related structures

Kyurangers or related terms

character followed by action

No Comments

No Known Activations