INDEX

Explanations

attends to tokens preceded by the specific example "e.g." from various tokens that can provide further clarification or examples

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-att/layer_0/width_16k/average_l0_61

Prompts (Dashboard)

16,384 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

0:0.02

1:0.01

2:0.01

3:0.06

4:0.01

5:0.04

6:0.02

7:0.03

8:0.01

9:0.01

10:0.01

11:0.01

12:0.03

13:0.61

14:0.02

15:0.01

Negative Logits

<_>

-0.07

-0.06

queryInterface

-0.06

}}_{\

-0.06

))=\

-0.06

<bos>

-0.06

}{@

-0.05

POSITIVE LOGITS

Personensuche

0.56

<unused99>

0.52

 Geiſt

0.50

оргаш

0.48

 ſuo

0.48

𐑥

0.48

 wiſſen

0.47

 ſua

0.47

ꩻ

0.47

 ſelbſt

0.47

Activations Density 0.296%

attends to tokens preceded by the specific example "e.g." from various tokens that can provide further clarification or examples

No Comments

No Known Activations