INDEX

Explanations

attends to tokens related to organizational roles or titles from named individuals or departments in a context of academia and governance

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

0:0.06

1:0.08

2:0.09

3:0.12

4:0.10

5:0.05

6:0.32

7:0.15

Negative Logits

modelBuilder

-0.30

걍

-0.29

 грамм

-0.29

luß

-0.28

 fucking

-0.28

atoires

-0.28

Whilst

-0.27

 Frankfurt

-0.27

ництво

-0.27

 FUCKING

-0.26

POSITIVE LOGITS

 nahilalakip

0.38

 विश्वसनीयता

0.37

archiviato

0.35

—

0.35

Diwedd

0.35

DockStyle

0.33

 nobis

0.32

StoryboardSegue

0.30

 Lisp

0.29

Personendaten

0.29

Activations Density 0.119%

attends to tokens related to organizational roles or titles from named individuals or departments in a context of academia and governance

No Comments

No Known Activations