INDEX

Explanations

attends to former-related tokens from various tokens that follow

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 10-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_10/width_16k/average_l0_70

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.10.attn.hook_z

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.11

1:0.12

2:0.13

3:0.04

4:0.03

5:0.02

6:0.07

7:0.45

Negative Logits

 MenuView

-0.32

bootstrapcdn

-0.31

Itoa

-0.31

Portale

-0.30

AccessFile

-0.29

دانشنامهٔ

-0.29

 expiration

-0.28

RAINT

-0.28

AddHtmlAttribute

-0.28

شهاد

-0.28

POSITIVE LOGITS

 magique

0.27

lrrr

0.27

 Wikimedijinoj

0.26

ertale

0.26

levis

0.25

 chimique

0.25

幸いです

0.24

تقاوى

0.24

lrrrr

0.24

äiv

0.23

Activations Density 0.076%

attends to former-related tokens from various tokens that follow

No Comments

No Known Activations