INDEX

Explanations

attends to phrases that include the word “from” from adjacent tokens that end with brackets

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

1:0.12

2:0.09

3:0.06

4:0.05

5:0.03

6:0.06

7:0.43

Negative Logits

AndEndTag

-0.46

BeginContext

-0.43

 consultato

-0.41

__*/

-0.38

 linkovi

-0.37

 MotionEvent

-0.37

RectangleBorder

-0.36

openConnection

-0.36

 Wikispecies

-0.35

ctrons

-0.35

POSITIVE LOGITS

RegressionTest

0.30

0.27

ynnwys

0.27

onicus

0.25

ykite

0.25

 respectively

0.24

}$.

0.23

Ἁ

0.23

absch

0.23

nical

0.23

Activations Density 0.005%

attends to phrases that include the word “from” from adjacent tokens that end with brackets

No Comments

No Known Activations