INDEX

Explanations

attends to tokens marked by specific symbols or labels from numerical values

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

1:0.16

2:0.10

3:0.06

4:0.08

5:0.04

6:0.07

7:0.32

Negative Logits

providedIn

-0.33

EDEFAULT

-0.32

ScopeManager

-0.32

 ddelweddau

-0.30

 grises

-0.29

 beginnetje

-0.29

MessageTagHelper

-0.29

 HttpNotFound

-0.28

scriptId

-0.28

 ModelExpression

-0.28

POSITIVE LOGITS

@@@@@@@@

0.28

 referrerpolicy

0.26

許

0.23

ınca

0.23

 houſe

0.22

 becauſe

0.22

 Inactive

0.22

 perſon

0.22

المكان

0.21

inactive

0.21

Activations Density 0.005%

attends to tokens marked by specific symbols or labels from numerical values

No Comments

No Known Activations