INDEX

Explanations

attends to the token "under" from tokens that specify contexts or conditions associated with it

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

0:0.05

1:0.02

2:0.02

3:0.02

4:0.11

5:0.70

6:0.02

7:0.03

Negative Logits

 Anſ

-0.53

 Shakspeare

-0.52

 AppColors

-0.52

 photolibrary

-0.51

 Monfieur

-0.51

 Shaksp

-0.50

 Reſ

-0.47

 ―――――

-0.46

 Theſe

-0.45

 Cæsar

-0.44

POSITIVE LOGITS

the

0.41

↵

0.41

“

0.38

is

0.35

0.34

he

0.32

 also

0.31

«

0.31

Activations Density 0.083%

attends to the token "under" from tokens that specify contexts or conditions associated with it

No Comments

No Known Activations