INDEX

Explanations

attends to specific technical or scientific terms marked by tokens from similar categories

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

0:0.04

1:0.10

2:0.06

3:0.05

4:0.21

5:0.44

6:0.03

7:0.04

Negative Logits

 mères

-0.27

 presidencial

-0.24

 başına

-0.24

 saját

-0.23

 leçons

-0.22

 preuves

-0.22

 helicópter

-0.22

meren

-0.21

 vált

-0.21

POSITIVE LOGITS

 CreateTagHelper

0.56

tagHelperRunner

0.54

LLocation

0.53

 imageNamed

0.49

principalColumn

0.48

下载附件

0.48

^(@)

0.47

 OnTriggerEnter

0.45

 MonoBehaviour

0.45

Hochspringen

0.44

Activations Density 1.453%

attends to specific technical or scientific terms marked by tokens from similar categories

No Comments

No Known Activations