INDEX

Explanations

attends to the token "List" from programming token "Set."

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

0:0.21

1:0.20

2:0.25

3:0.05

4:0.06

5:0.04

6:0.06

7:0.08

Negative Logits

ArgsConstructor

-0.35

//

-0.31

 سكانية

-0.31

 numberOfRows

-0.29

 kysy

-0.29

<h3>

-0.28

الحياه

-0.28

PyExc

-0.28

✭✭

-0.28

AndroidJUnit

-0.28

POSITIVE LOGITS

 /\.(

0.35

contentLoaded

0.31

 ErrIntOverflow

0.30

devise

0.30

 transfieras

0.29

CodeAttribute

0.29

цездатний

0.28

 ddelweddau

0.28

 classNames

0.28

Autoritní

0.27

Activations Density 0.008%

attends to the token "List" from programming token "Set."

No Comments

No Known Activations