INDEX

Explanations

attends to tokens referring to types of items or categories from the corresponding tokens that appear later

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.04

1:0.77

2:0.03

3:0.01

4:0.03

5:0.04

6:0.01

7:0.02

Negative Logits

-0.29

…

-0.29

<eos>

-0.26

-0.24

ms

-0.24

 lemn

-0.24

-0.23

-0.21

fs

-0.21

POSITIVE LOGITS

featureID

0.64

 AssemblyCulture

0.61

 tartalomajánló

0.60

SourceChecksum

0.54

NameInMap

0.53

ItemBackground

0.52

InjectAttribute

0.51

 &___

0.51

 تكبرها

0.50

Personendaten

0.50

Activations Density 7.877%

attends to tokens referring to types of items or categories from the corresponding tokens that appear later

No Comments

No Known Activations