INDEX

Explanations

attends to the token "was" from specific historical tokens relating to the Barracuda model

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

0:0.09

1:0.11

2:0.10

3:0.13

4:0.10

5:0.06

6:0.20

7:0.18

Negative Logits

put

-0.25

Put

-0.25

ftagPool

-0.24

 كومونز

-0.24

put

-0.23

leb

-0.23

 ویکی‌پدی

-0.23

Put

-0.23

bilt

-0.23

ContentAlignment

-0.23

POSITIVE LOGITS

verwijspagina

0.40

 Мексичка

0.38

########.

0.37

FTFY

0.36



0.35

."</

0.34

 Administrativna

0.33

+#+#

0.33

erequisite

0.32

\]

0.32

Activations Density 0.216%

attends to the token "was" from specific historical tokens relating to the Barracuda model

No Comments

No Known Activations