INDEX

Explanations

phrases that indicate high-quality or elite attributes in various contexts

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

preceding a description of quality or rank

np_acts-logits-general · gemini-2.0-flash

high quality or advanced subjects

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

Hochspringen

-0.93

SequentialGroup

-0.82

 فريبيس

-0.81

complexContent

-0.67

parsedMessage

-0.64

contentLoaded

-0.63

RetentionPolicy

-0.61

 resourceCulture

-0.58

RenderAtEndOf

-0.58

pielen

-0.56

POSITIVE LOGITS

 quality

0.72

 himſelf

0.67

 grade

0.65

 themſelves

0.65

 status

0.64

 houſe

0.62

íssima

0.62

 ―――――

0.62

 level

0.61

 pleaſure

0.61

Activations Density 0.219%

phrases that indicate high-quality or elite attributes in various contexts

preceding a description of quality or rank

high quality or advanced subjects

No Comments

No Known Activations