INDEX

Explanations

attends to tokens indicating a "to" relationship from corresponding "from" tokens

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

0:0.12

1:0.19

2:0.36

3:0.10

4:0.06

5:0.02

6:0.04

7:0.07

Negative Logits

 disambiguazione

-0.39

 magari

-0.38

//

-0.36

 perhaps

-0.32

sizeCache

-0.31

 further

-0.30

AntiForgeryToken

-0.29

EndContext

-0.29

 maybe

-0.29

perhaps

-0.29

POSITIVE LOGITS

 gynhyrchwyd

0.35

aarrggbb

0.33

ativna

0.32

 HasFactory

0.30

 Chwiliwch

0.30

@+

0.30

Население

0.29

Hauptartikel

0.28

Referencies

0.28

 ventricle

0.28

Activations Density 0.088%

attends to tokens indicating a "to" relationship from corresponding "from" tokens

No Comments

No Known Activations

attends to tokens indicating a "to" relationship from corresponding "from" tokens

No Comments

No Known Activations