INDEX

Explanations

attends to action-related tokens from resulting state 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.29

1:0.28

2:0.10

3:0.05

4:0.06

5:0.03

6:0.06

7:0.09

Negative Logits

URLException

-0.42

endpush

-0.35

'},

-0.33

󠁿

-0.32

✨:

-0.32

onViewCreated

-0.32

Geplaatst

-0.29

']);

-0.28

lapsingToolbar

-0.28

Portail

-0.28

POSITIVE LOGITS

 tanı

0.30

انيف

0.30

انجليز

0.28

awtextra

0.27

 Marcell

0.25

angliski

0.24

 apor

0.24

нда

0.24

Ế

0.24

 sfondo

0.24

Activations Density 0.687%

attends to action-related tokens from resulting state tokens

No Comments

No Known Activations