INDEX

Explanations

historical events and dates

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

invented after 1920

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

 queſta

-0.68

ſelves

-0.66

ſelf

-0.65

 houſe

-0.65

 poffible

-0.62

setVerticalGroup

-0.62

 leſs

-0.61

 ſtate

-0.61

$_"

-0.60

 leaſt

-0.60

POSITIVE LOGITS

 kick

0.35

pan

0.34

 down

0.33

kick

0.32

 panning

0.32

 trip

0.31

 period

0.30

 kicking

0.30

BeginInit

0.29

esehen

0.29

Activations Density 0.051%

historical events and dates

invented after 1920

No Comments

No Known Activations