INDEX

Explanations

attends to Boolean state indicators from the negated Boolean state indicators

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

0:0.11

1:0.12

2:0.16

3:0.04

4:0.04

5:0.04

6:0.07

7:0.39

Negative Logits

SCRIBE

-0.31

 Lieberman

-0.27

 Sarm

-0.26

kosi

-0.26

Her

-0.26

 acquittal

-0.24

 Krie

-0.24

Her

-0.24

ethene

-0.23

 spurious

-0.23

POSITIVE LOGITS

WithIOException

0.41

uxxxx

0.39

Personensuche

0.39

 للاسماء

0.36

 AssemblyCompany

0.35

routeProvider

0.35

awaiter

0.34

########.

0.33

Hentet

0.33

irent

0.33

Activations Density 0.165%

attends to Boolean state indicators from the negated Boolean state indicators

No Comments

No Known Activations