INDEX

Explanations

attends to punctuation marks from tokens with conversational phrases or questions

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

0:0.18

1:0.22

2:0.20

3:0.08

4:0.06

5:0.03

6:0.07

7:0.11

Negative Logits

তথ্যসূত্র

-0.55

 unknownFields

-0.47

VY

-0.39

 StatefulWidget

-0.38

DPI

-0.38

 StatelessWidget

-0.38

 Alder

-0.37

 QName

-0.37

 LaLiga

-0.36

||}

-0.36

POSITIVE LOGITS

tawesome

0.42

addObject

0.37

ंदीखरीदारी

0.36

vensko

0.35

rafat

0.34

minecraft

0.33

lastIndexOf

0.33

enken

0.33

eders

0.32

 time

0.31

Activations Density 0.035%

attends to punctuation marks from tokens with conversational phrases or questions

No Comments

No Known Activations