INDEX

Explanations

attends to specific tokens from the second token in phrases that typically structure definitions or explanations

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

0:0.14

1:0.17

2:0.12

3:0.10

4:0.13

5:0.04

6:0.11

7:0.17

Negative Logits

gridx

-0.49

########.

-0.48

SourceChecksum

-0.46

amaño

-0.45

 loginUser

-0.45

̀u

-0.44

queryInterface

-0.43

])->

-0.43

fören

-0.43

")));

-0.42

POSITIVE LOGITS

ISD

0.36

 Butterworth

0.36

gester

0.36

 Mehl

0.36

ISD

0.36

》

0.36

 paraître

0.35

 Baik

0.35

 Broadway

0.34

din

0.34

Activations Density 31.339%

attends to specific tokens from the second token in phrases that typically structure definitions or explanations

No Comments

No Known Activations