INDEX

Explanations

attends to function-like tokens from tokens that are self-referential in nature

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

0:0.12

1:0.12

2:0.11

3:0.12

4:0.07

5:0.03

6:0.09

7:0.30

Negative Logits

 poches

-0.31

 Butterworth

-0.29

camy

-0.28

nasional

-0.28

InputBorder

-0.28

Separator

-0.28

onucleotide

-0.28

 remplissage

-0.28

HIND

-0.27

cheibe

-0.27

POSITIVE LOGITS

featureID

0.34

 mxArray

0.29

Viki

0.28

kháu

0.28

NewGuid

0.26

']>;

0.25

 الرياضيه

0.25

rungsseite

0.25

Šaltiniai

0.25

'{@

0.25

Activations Density 0.144%

attends to function-like tokens from tokens that are self-referential in nature

No Comments

No Known Activations