INDEX

Explanations

attends to interrogative tokens from follow-up clarifying 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

3:0.11

4:0.10

5:0.06

6:0.20

7:0.22

Negative Logits

writeField

-0.27

 TextEditing

-0.26

 nationales

-0.25

PointerException

-0.25

 étrangère

-0.25

 viagens

-0.25

endaft

-0.25

 língua

-0.25

IsMutable

-0.24

 hauts

-0.24

POSITIVE LOGITS

Hochspringen

0.32

Try

0.30

adaptiveStyles

0.28

-->

0.27

helin

0.27

 forti

0.26

 myſelf

0.26

migrationBuilder

0.26

 Chio

0.26

par

0.26

Activations Density 0.020%

attends to interrogative tokens from follow-up clarifying tokens

No Comments

No Known Activations