INDEX

Explanations

attends to concluding tokens from preceding 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

0:0.05

1:0.07

2:0.07

3:0.12

4:0.10

5:0.03

6:0.37

7:0.15

Negative Logits

 tartalomajánló

-0.36

bbene

-0.35

ToMany

-0.33

basicConfig

-0.33

UnusedPrivate

-0.32

Chham

-0.32

nál

-0.31

CppMethod

-0.31

WebServlet

-0.31

 tarko

-0.31

POSITIVE LOGITS

bootstrapcdn

0.30

Hrsg

0.30

Wal

0.29

 oprot

0.27

erequisite

0.27

principalColumn

0.26

Filmographie

0.25

AnchorStyles

0.25

])));

0.24

FundMe

0.24

Activations Density 0.397%

attends to concluding tokens from preceding tokens

No Comments

No Known Activations