INDEX

Explanations

attends to varied token types followed by asterisked tokens from sequences starting with asterisks

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.10

1:0.15

2:0.14

3:0.11

4:0.13

5:0.07

6:0.13

7:0.14

Negative Logits

]")]

-0.34

mogat

-0.32

lucas

-0.32

 Zeitung

-0.30

']]

-0.30

intah

-0.29

rags

-0.29

%)$

-0.29

στη

-0.29

PreferredItem

-0.29

POSITIVE LOGITS

AccessorTable

0.42

tonsoft

0.40

MemoryWarning

0.40

 ویکی‌پدیا

0.40

LookAnd

0.38

 protoimpl

0.38

 Exactos

0.36

 HasFactory

0.34

Personensuche

0.34

Ech

0.34

Activations Density 0.024%

attends to varied token types followed by asterisked tokens from sequences starting with asterisks

No Comments

No Known Activations