INDEX

Explanations

attends to note-related tokens from various contextual explanations or elaborations

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

2:0.07

3:0.12

4:0.09

5:0.02

6:0.40

7:0.14

Negative Logits

 GenerationType

-0.30

 wireType

-0.28

tidaknya

-0.27

ngdoc

-0.27

matorium

-0.26

 Engine

-0.26

Engine

-0.26

reale

-0.25

vented

-0.25

FFF

-0.25

POSITIVE LOGITS

 CanadaChoose

0.37

 professionali

0.33

invokeLater

0.33

ftagPool

0.33

 onlyOwner

0.32

?>>

0.32

 bianchi

0.32

 anteriore

0.31

 indiri

0.31

webElementGuid

0.31

Activations Density 0.665%

attends to note-related tokens from various contextual explanations or elaborations

No Comments

No Known Activations