INDEX

Explanations

attends to tokens marked as significant or important terms, possibly from other tokens that provide additional context or details

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

0:0.10

1:0.13

2:0.12

3:0.17

4:0.12

5:0.02

6:0.20

7:0.11

Negative Logits

RenderAtEndOf

-0.45

 autorytatywna

-0.40

حياتها

-0.35

FormTagHelper

-0.34

Efq

-0.34

 HasFactory

-0.33

 saites

-0.33

 lenker

-0.32

 Majefty

-0.32

 Shakspeare

-0.32

POSITIVE LOGITS

heen

0.28

 dalamnya

0.27

 Fore

0.26

(")");

0.24

Hos

0.24

ccedil

0.24

 BeautifulSoup

0.23

fun

0.23

minecraft

0.23

oglobulin

0.23

Activations Density 0.000%

attends to tokens marked as significant or important terms, possibly from other tokens that provide additional context or details

No Comments

No Known Activations