INDEX

Explanations

attends to subject-related tokens that express a state of being from corresponding descriptors related to identity or existence

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

3:0.10

4:0.10

5:0.03

6:0.34

7:0.20

Negative Logits

 myſelf

-0.47

 houſe

-0.45

 reaſon

-0.45

 themſelves

-0.44

 ſtate

-0.42

 purpoſe

-0.42

 fuper

-0.42

 perfons

-0.41

 neceff

-0.41

ſelf

-0.40

POSITIVE LOGITS

 '\\;'

0.42

]--;

0.31

usercontent

0.31

nebago

0.29

enment

0.29

addContainerGap

0.29

0.28

ins

0.28

ensement

0.28

 Schult

0.27

Activations Density 0.507%

attends to subject-related tokens that express a state of being from corresponding descriptors related to identity or existence

No Comments

No Known Activations