INDEX

Explanations

attends to tokens marked with specific keywords or object names from accompanying tokens

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

0:0.07

1:0.07

2:0.54

3:0.09

4:0.05

5:0.03

6:0.04

7:0.06

Negative Logits

MLLoader

-0.33

Revenir

-0.32

ItemIndex

-0.31

cig

-0.30

phy

-0.30

Redu

-0.30

VIAF

-0.30

moduleId

-0.29

zoni

-0.29

few

-0.29

POSITIVE LOGITS

 BoxDecoration

0.37

ImageContext

0.35

 виправивши

0.31

^(@)

0.31

InputTagHelper

0.31

antaranya

0.31

发表于

0.30

 insuffisamment

0.30

pozdrawiam

0.30

Dimensional

0.29

Activations Density 0.054%

attends to tokens marked with specific keywords or object names from accompanying tokens

No Comments

No Known Activations