INDEX

Explanations

attends to the first token and multiple variations of it from the corresponding token attending to itself

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_0/width_16k/average_l0_104

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.0.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

0:0.01

1:0.03

2:0.53

3:0.16

4:0.05

5:0.06

6:0.05

7:0.06

Negative Logits

 back

-0.28

aarrggbb

-0.27

migrationBuilder

-0.26

dra

-0.23

 AssemblyTitle

-0.23

 коли

-0.23

 kinh

-0.23

setPointSize

-0.22

 invokingState

-0.22

TabIndex

-0.22

POSITIVE LOGITS

دانشنامهٔ

0.46

HostException

0.39

clusal

0.36

WriteTagHelper

0.35

󠁣

0.35

 дописавши

0.34

 }}$}

0.34

)}</

0.33

AsUp

0.32

IVEREF

0.32

Activations Density 4.290%

attends to the first token and multiple variations of it from the corresponding token attending to itself

No Comments

No Known Activations