INDEX

Explanations

attends to JSON-related tokens from tokens framed within the context of JSON processing

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

0:0.51

1:0.02

2:0.03

3:0.04

4:0.19

5:0.11

6:0.02

7:0.04

Negative Logits

rxjs

-0.40

TemporalType

-0.38

 alimentaires

-0.38

 discorso

-0.38

 clearColor

-0.38

 âgées

-0.37

 BrowserRouter

-0.37

 commerciales

-0.37

OGND

-0.35

 vérit

-0.34

POSITIVE LOGITS

ardless

0.27

OrWhiteSpace

0.27

ContentAsync

0.27

 Völker

0.27

 capture

0.26

</b>

0.25

iprot

0.25

 Kanada

0.25

 igni

0.25

WebServlet

0.25

Activations Density 0.014%

attends to JSON-related tokens from tokens framed within the context of JSON processing

No Comments

No Known Activations