INDEX

Explanations

specific references to time or events

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

week day year

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

 mieux

-0.34

DllImport

-0.31

AZIONI

-0.31

wirtschaft

-0.30

lexikon

-0.28

gæ

-0.27

 числе

-0.27

RunWith

-0.27

 behulp

-0.27

までに

-0.26

POSITIVE LOGITS

 week

0.77

 year

0.74

 YEAR

0.69

day

0.69

wiliwch

0.69

 years

0.66

 late

0.66

 tahun

0.65

 WEEK

0.65

ScopeManager

0.64

Activations Density 0.090%

specific references to time or events

week day year

No Comments

No Known Activations