INDEX

Explanations

attends to the numeric value and operator tokens from the related 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.04

1:0.09

2:0.05

3:0.03

4:0.21

5:0.46

6:0.03

7:0.04

Negative Logits

stdc

-0.29

CppCodeGen

-0.26

MathML

-0.25

ToFit

-0.24

 للاسماء

-0.23

 <<<<<<<<<<<<<<

-0.23

 initComponents

-0.21

RuntimeObject

-0.21

audiovisuel

-0.21

MEMORANDUM

-0.21

POSITIVE LOGITS

 themſelves

0.31

 fourrure

0.31

 avoient

0.31

 myſelf

0.31

 vierge

0.31

 guerres

0.30

 espagne

0.30

 altså

0.29

 poffible

0.28

 neuve

0.28

Activations Density 0.493%

attends to the numeric value and operator tokens from the related tokens

No Comments

No Known Activations