INDEX

Explanations

attends to specific tokens from lists and indices in data structures

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

1:0.17

2:0.15

3:0.12

4:0.13

5:0.06

6:0.09

7:0.11

Negative Logits

 استنادى

-0.32

jší

-0.27

ling

-0.26

tetrahydro

-0.26

то

-0.26

gantian

-0.25

стами

-0.25

mb

-0.25

īb

-0.25

lijk

-0.24

POSITIVE LOGITS

endphp

0.51

iastes

0.47

])));

0.47

 StatelessWidget

0.47

 فريبيس

0.44

 سكانية

0.43

 NSCoder

0.41

 CreateTagHelper

0.40

ftagPool

0.40

']]

0.39

Activations Density 0.076%

attends to specific tokens from lists and indices in data structures

No Comments

No Known Activations