INDEX

Explanations

attends to "left" tokens from "left" tokens in a 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

1:0.18

2:0.04

3:0.04

4:0.13

5:0.39

6:0.05

7:0.05

Negative Logits

 autorytatywna

-0.36

ⓧ

-0.33

 ſtate

-0.32

Chham

-0.31

 purpoſe

-0.31

 дописавши

-0.31

tableFuture

-0.31

 ―――――

-0.31

FunctionFlags

-0.30

 pleaſure

-0.30

POSITIVE LOGITS

 blessés

0.27

 indépendant

0.25

 nomme

0.25

 privées

0.24

 privado

0.24

ähkö

0.24

otomy

0.23

endiri

0.23

 Private

0.23

 scolaire

0.23

Activations Density 0.166%

attends to "left" tokens from "left" tokens in a sequence

No Comments

No Known Activations