INDEX

Explanations

attends to matrix-related tokens from mathematical expressions or definitions in text

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.14

1:0.17

2:0.16

3:0.10

4:0.11

5:0.07

6:0.07

7:0.15

Negative Logits

TagMode

-0.31

twimg

-0.30

AVR

-0.28

 EconPapers

-0.28

publicain

-0.27

extAlignment

-0.26

 CURIAM

-0.26

 strike

-0.26

ISupport

-0.26

[]"

-0.26

POSITIVE LOGITS

nloa

0.21

DoubleQuotes

0.18

Facades

0.18

منابع

0.18

 SUDOC

0.18

SequentialGroup

0.18

postValue

0.17

 насељу

0.17

 ngoại

0.17

 Interest

0.17

Activations Density 0.526%

attends to matrix-related tokens from mathematical expressions or definitions in text

No Comments

No Known Activations