INDEX

Explanations

attends to index-related tokens from comparison-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

0:0.40

1:0.20

2:0.09

3:0.04

4:0.05

5:0.08

6:0.03

7:0.06

Negative Logits

errHandler

-0.35

HasForeignKey

-0.32

 Sinks

-0.32

 Danemark

-0.32

 évident

-0.31

úrese

-0.31

jand

-0.30

 heureux

-0.30

InstanceState

-0.30

Curator

-0.30

POSITIVE LOGITS

marle

0.39

########.

0.39



0.37

lawayo

0.35

}());

0.34

 Bourgoin

0.33

SBATCH

0.30

ignements

0.29

ffindor

0.28

udesta

0.28

Activations Density 0.498%

attends to index-related tokens from comparison-related tokens

No Comments

No Known Activations