INDEX

Explanations

attends to tokens marked as non-word characters, primarily punctuation and symbols, from tokens that tend to indicate structure or categories 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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.06

1:0.07

2:0.08

3:0.38

4:0.14

5:0.04

6:0.11

7:0.07

Negative Logits

 الرياضيه

-0.39

JsonHelper

-0.35

KURZBESCHREIBUNG

-0.33

 Yandex

-0.30

 Signalez

-0.30

 münd

-0.30

 Tham

-0.30

 repro

-0.29

readlines

-0.29

 Clim

-0.29

POSITIVE LOGITS

 محفوظة

0.38

ArgsConstructor

0.32

expanded

0.32

 للمعارف

0.32

Становништво

0.28

tagHelper

0.27

 expanded

0.26

IndentedString

0.26

Hauptartikel

0.26

 aidant

0.26

Activations Density 0.993%

attends to tokens marked as non-word characters, primarily punctuation and symbols, from tokens that tend to indicate structure or categories in text

No Comments

No Known Activations