INDEX

Explanations

attends to tokens in the context of mechanical or technical components from their respective counterpart 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

0:0.06

1:0.13

2:0.07

3:0.07

4:0.15

5:0.35

6:0.05

7:0.08

Negative Logits

by

-0.29

 khảo

-0.28

prepareStatement

-0.26

 egli

-0.26

execSQL

-0.26

 faits

-0.25

ülerin

-0.25

pr

-0.25

τώ

-0.24

 متعلقه

-0.24

POSITIVE LOGITS

ImageContext

0.35

 سكانية

0.30

 насељу

0.30

ppuden

0.29

daß

0.28

mybatisplus

0.27

MZ

0.27

RectangleBorder

0.27

gebob

0.27

SceneManagement

0.27

Activations Density 0.361%

attends to tokens in the context of mechanical or technical components from their respective counterpart tokens

No Comments

No Known Activations