INDEX

Explanations

attends to numerical tokens from reference-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

0:0.08

1:0.11

2:0.17

3:0.08

4:0.20

5:0.16

6:0.07

7:0.08

Negative Logits

ChildScrollView

-0.33

aarrggbb

-0.32

 للاسماء

-0.31

onee

-0.28

 tege

-0.28

 AssemblyCompany

-0.28

evos

-0.28

[++

-0.27

Rhestr

-0.27

发表于

-0.27

POSITIVE LOGITS

RenderAtEndOf

0.30

AndEndTag

0.28

 verze

0.27

konstru

0.27

 înc

0.27

 <<<<<<<<<<<<<<

0.26

יסטור

0.26

 člá

0.25

 Pandora

0.25

LayoutStyle

0.24

Activations Density 0.051%

attends to numerical tokens from reference-related tokens

No Comments

No Known Activations