INDEX

Explanations

attends to temporal or numerical variances from corresponding tokens indicating a previous state or amount

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

0:0.10

1:0.11

2:0.09

3:0.04

4:0.03

5:0.01

6:0.06

7:0.52

Negative Logits

-0.27

 again

-0.27

yd

-0.26

知

-0.24

 bene

-0.24

again

-0.24

äť

-0.24

πό

-0.23

ié

-0.23

ContentAlignment

-0.23

POSITIVE LOGITS

InjectAttribute

0.43

 itſelf

0.42

/**

0.42

 Reſ

0.40

 defStyleAttr

0.39

 myſelf

0.39

ODS

0.38

 ་་

0.37

GOTREF

0.37

ſelves

0.37

Activations Density 0.306%

attends to temporal or numerical variances from corresponding tokens indicating a previous state or amount

No Comments

No Known Activations