INDEX

Explanations

attends to instances of the token "writer" from instances of the token "writer" appearing later in the sequence

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

0:0.07

1:0.09

2:0.06

3:0.04

4:0.21

5:0.39

6:0.05

7:0.05

Negative Logits

 élé

-0.28

 چیست

-0.26

kano

-0.26

ruitment

-0.24

:''

-0.24

 μή

-0.23

 atra

-0.23

 fermeture

-0.23

sû

-0.23

uker

-0.23

POSITIVE LOGITS

 GlobalKey

0.36

ChildScrollView

0.33

 الحره

0.32

 ImGui

0.32

;->

0.31

RenderAtEndOf

0.31

 heapq

0.30

 AUSSI

0.29

gouv

0.29

CDCl

0.29

Activations Density 0.082%

attends to instances of the token "writer" from instances of the token "writer" appearing later in the sequence

No Comments

No Known Activations