INDEX

Explanations

attends to numerical tokens from identical numerical tokens

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

0:0.11

1:0.17

2:0.22

3:0.06

4:0.12

5:0.17

6:0.04

7:0.07

Negative Logits

NameInMap

-0.41

 Мексичка

-0.39

*/].

-0.32

 &___

-0.32

IFICATE

-0.30

enumii

-0.29

BeginInit

-0.29

 виправивши

-0.29

werfer

-0.29

 EconPapers

-0.29

POSITIVE LOGITS

ScopeManager

0.31

 airs

0.30

elfth

0.29

 invokingState

0.28

 חיצוניים

0.28

SpringBootTest

0.26

IMPORTED

0.26

Traveler

0.26

DebuggerNonUser

0.26

Copy

0.25

Activations Density 0.027%

attends to numerical tokens from identical numerical tokens

No Comments

No Known Activations

attends to numerical tokens from identical numerical tokens

No Comments

No Known Activations