INDEX

Explanations

attends to programming-related keywords from neighboring tokens

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

0:0.11

1:0.35

2:0.10

3:0.05

4:0.08

5:0.04

6:0.14

7:0.10

Negative Logits

لينكات

-0.32

didSet

-0.32

atellite

-0.31

 BitmapFactory

-0.31

msgSender

-0.30

SequentialGroup

-0.30

 ActivatedRoute

-0.29

 shutil

-0.29

shutil

-0.29

<caption>

-0.28

POSITIVE LOGITS

 purpoſe

0.47

 houſe

0.42

 greateſt

0.40

 ſever

0.40

 pleaſure

0.40

 cauſe

0.40

 reaſon

0.39

 ſtate

0.39

uxxxx

0.39

 Houſe

0.39

Activations Density 0.035%

attends to programming-related keywords from neighboring tokens

No Comments

No Known Activations