INDEX

Explanations

attends to words that serve as general descriptors or qualifiers from words that specify or define a concept

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

0:0.13

1:0.19

2:0.17

3:0.06

4:0.07

5:0.02

6:0.06

7:0.25

Negative Logits

 оригіналу

-0.26

ind

-0.24

się

-0.24

 Baker

-0.24

bourhood

-0.23

al

-0.23

ticles

-0.22

inck

-0.22

falda

-0.22

az

-0.22

POSITIVE LOGITS

TagMode

0.36

 ་་

0.34

 Torr

0.33

BufferException

0.31

^(@)

0.31

ujudkan

0.30

RegistryLite

0.30

зулта

0.29

DockStyle

0.29

 ThemeData

0.28

Activations Density 0.175%

attends to words that serve as general descriptors or qualifiers from words that specify or define a concept

No Comments

No Known Activations