INDEX

Explanations

attends to transformation-related tokens from class-related 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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.08

1:0.10

2:0.09

3:0.16

4:0.13

5:0.08

6:0.18

7:0.14

Negative Logits

opardy

-0.24

__(/*!

-0.24

 erfolgre

-0.23

builtin

-0.23

 pesanan

-0.23

logna

-0.22

 lệnh

-0.22

UVWXYZ

-0.21

 gesetz

-0.21

 ansatte

-0.21

POSITIVE LOGITS

InjectAttribute

0.39

 ThemeData

0.36

UnsafeEnabled

0.36

 становника

0.35

 محفوظة

0.35

MethodManager

0.34

PerformLayout

0.34

 שוליים

0.31

afficheront

0.31

 صوتيه

0.31

Activations Density 0.036%

attends to transformation-related tokens from class-related tokens

No Comments

No Known Activations