INDEX

Explanations

attends to various instances of "AN" from corresponding tokens that follow it in the sequence

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_0/width_16k/average_l0_104

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.0.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

0:0.59

1:0.01

2:0.03

3:0.02

4:0.17

5:0.04

6:0.02

7:0.08

Negative Logits

nowa

-0.36

sightly

-0.35

himovic

-0.35

cie

-0.34

goe

-0.34

Compat

-0.32

ь

-0.32

JAH

-0.31

 otomatig

-0.31

andler

-0.31

POSITIVE LOGITS

 purpoſe

0.43

 Monfieur

0.41

 Majefty

0.39

 becauſe

0.38

 Jefus

0.38

 myſelf

0.38

 poffible

0.37

Хьажоргаш

0.37

 fubject

0.37

MENAFN

0.36

Activations Density 0.019%

attends to various instances of "AN" from corresponding tokens that follow it in the sequence

No Comments

No Known Activations