INDEX

Explanations

attends to variable tokens associated with "h" from tokens that are represented as unknown or with a specific symbol

oai_attention-head · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 0-gemmascope-att-16k

Configuration

google/gemma-scope-2b-pt-att/layer_0/width_16k/average_l0_104

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.0.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

0:0.76

1:0.01

2:0.01

3:0.01

4:0.11

5:0.01

6:0.02

7:0.03

Negative Logits

 ſhe

-0.45

 faſt

-0.43

 ſte

-0.43

 uſed

-0.42

 ſtand

-0.42

 deſt

-0.41

ImageField

-0.41

 photolibrary

-0.41

 ſta

-0.41

 reaſon

-0.41

POSITIVE LOGITS

bootstrapcdn

0.30

 Inter

0.29

 inter

0.28

 Senin

0.28

 which

0.27

he

0.27

ho

0.27

webElementXpaths

0.26

án

0.26

invoke

0.26

Activations Density 0.018%

attends to variable tokens associated with "h" from tokens that are represented as unknown or with a specific symbol

No Comments

No Known Activations