INDEX

Explanations

repeated mentions of time or past events, particularly in relation to performance or outcomes

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

previous or prior events

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

 Administrativna

-0.76

sizeCache

-0.73

Personensuche

-0.73

 EconPapers

-0.63

ImageContext

-0.62

帖最后由

-0.60

RUnlock

-0.60

 ſche

-0.59

DockStyle

-0.59

styleUrls

-0.59

POSITIVE LOGITS

 previous

0.69

 earlier

0.57

previous

0.56

Previous

0.47

 précédent

0.46

 предыду

0.45

 Previous

0.44

 précédente

0.44

 past

0.43

 прошлого

0.42

Activations Density 0.090%

repeated mentions of time or past events, particularly in relation to performance or outcomes

previous or prior events

No Comments

No Known Activations