INDEX

Explanations

attends to usage-related tokens from related specifications or components

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

0:0.08

1:0.11

2:0.10

3:0.13

4:0.11

5:0.04

6:0.21

7:0.18

Negative Logits

\{\\

-0.33

ContentAsync

-0.32

 >=",

-0.32

 AssemblyCompany

-0.31

InvalidProtocol

-0.31

◄

-0.30

Sut

-0.30

AssignableFrom

-0.30

MarshalTo

-0.30

Personensuche

-0.30

POSITIVE LOGITS

Ecotoxicity

0.27

PageModule

0.26

 AIAA

0.24

vestres

0.23

ForValue

0.23

 Minn

0.23

ibos

0.23

IDAS

0.23

 CYTHON

0.23

Географиясе

0.23

Activations Density 0.074%

attends to usage-related tokens from related specifications or components

No Comments

No Known Activations