Skip to main content

Arkadia AI Data Format (AID) - A versatile data serialization format optimized for AI applications.

This project has been archived.

The maintainers of this project have marked this project as archived. No new releases are expected.

Project description

ARKADIA AI.DATA-FORMAT

                             :SB@MMWBZSr,
                         ,irXX7ri, iaBWMMMMM@a:    :,
                    i2WMMMMMMMMMMMM@a:7ZZWMMMMM@7 :i,02,
                 ,7S7;i::i;72Z2S0@ZZWMW;r7iBWWM@M2 rM0BM;
                   ia@@08Z0B2:  :,i0r rBZ,: XSBMZM: 2MMMMi
                 iBZ7i:,   iXZZSi   i;  ia   ;:@22: ;M@MMS
                 ,:    ,, ,7:  ,Xi  ;r,   :   ,SS,  rBZ8MiX7,
             r8WZ: XW@0Z8WMZ: rX: :M2;BZ ;r:   ;,   ::Si;ii   :7
          :8MM@a,:@0:      XM2  , i@WZMZ 28MMMBr     , :ZW@a  iB@;
        ,Z0MMWZi BB  ,aZ8i      S7,,  ;SXZMMMMMMMa::r ,         7M8,
        :;MMZ:   8Z  :07Z7   @X XaSa,:8MMMM2ZMMMMMMM@MS   i2 ,BX :WW:
        iM0:  ,X,i8r        XM:,S7a; XZWMMMM2,8MMMMMWr8M2, iS,    ,B@:
        87  iB0:,,,XZX,  XW S@    ,;aZ00ZZ2SX7,:S8082,iM00, :i  ar ,WW,
        : :0MB@Zi,,      iM2 Sr :;Xriirr:  ZM8      ,XBM@X   :   ,: i@a
         rMMMW0X:  8Z     : ;a00Z7,,2BMMWBZXi  :,   iaWMMMMB7    :;  2Wi
        rMMBr     ZW ,i   rW7a00Z702 78@MMMMMMMM0@MMM@@@MMMMMM0r     iWS
       :MZ,  ,7: ;M2;Zii,:MrB,   2S07 ZZ8BZ,XB0WMMMMMMMMW280BWWWW8i   8Z
       S:  ,ZZ   XMZW:rr, Z;B:  ,ZXSi S2222XXX7ZBZ;:BMMM@X:   irr:    ZZ
          ;MZ;   rMMX   : ,27XZZS;S;     ii    ,i,ZMMMMMMMMM@7  ,8Z   aa
         ;MM0: ,  BMi  iS   ,irr;:;, :7,,Wa S8Z,,aXMZWBBMMMMMMS:MM8  ,ZX
        ,WMM0,2   :@,  0rr        ;@Br2,7S ::r7,ir;rr8ZZMMMMMMSiMMa  ;ai
        rMMM8a7    ,:  @0:  ,:    ,   ,         ,Z222X77@MW02i iZW: ,SX
        X@SM0@:        BB   ;r   :S     ,;7XS77; ,XZZZZ82r, :;i ,,  ;X:
        ;riMMM:  , r , ;0 ,:7Z  ;SS   ;     ,       ,:  iZ@MMMMM2
           8MM;  ; Z ;i ;, XrMi  B8   X7 , ,7        ,XZWMMMMMM@B
           ,WM2  7;87:8    XXMW, ,Bi ;7@i i Z     ,;2aa8MWWMMMM02
            ,B@, :ZaMiBX   :ZBMWr, ;  XB@iiXX2::     Xa7;80Z@0a :
              X8  r8WMaMr   iZ0MM0,    ,X@;;ZWX8       :S; 7:
               ,i  rBWMMMr   :X;WM@r      ir:0@W0
                    ,0@MMM2    : ;BMWi        7@MW:
                      :0MMMWi       ;aBS:       i8MZ:
                         XBMMW;                    ,;r,
                            iS0@8;

The High-Density, Token-Efficient Data Protocol for Large Language Models. Stop wasting context window on JSON braces. AI.DATA is a unified, schema-first data format designed specifically for AI understanding. It offers up to 25% token savings, faster parsing, and human-readable structure that LLMs love.


๐Ÿ“ฆ Installation

Get started immediately with pip:

pip install arkadia-ai-data-format

๐Ÿš€ Fast Example

Encoding to AI.DATA:

echo '{ "data": 2}' | aid enc - -c
# Output: <data:number>(2)

Decoding back to JSON:

echo '<data:number>(2)' | aid dec - -f json
# Output: { "data": 2 }

โšก Performance & Token Savings

Why switch? Because every token counts. AICD (Arkadia Compressed Data) consistently outperforms standard formats in both token efficiency.

BENCHMARK SUMMARY:


   JSON  โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–‘โ–‘โ–‘โ–‘     6921 tok   โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘     0.15 ms
   AICD  โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘     5416 tok   โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆ     4.40 ms
   AID   โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–‘โ–‘โ–‘โ–‘โ–‘โ–‘     6488 tok   โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–‘     4.29 ms
   TOON  โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆ     8198 tok   โ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–ˆโ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘โ–‘     2.36 ms


   FORMAT     TOKENS       TIME (Total)    AVG TIME/FILE   VS JSON
   ----------------------------------------------------------------------
   AICD       5416             4.40 ms        0.37 ms    -21.7%
   AID        6488             4.29 ms        0.36 ms    -6.3%
   JSON       6921             0.15 ms        0.01 ms    +0.0%
   TOON       8198             2.36 ms        0.20 ms    +18.5%


CONCLUSION: Switching to AICD saves 1505 tokens (21.7%) compared to JSON.

๐Ÿ›  CLI Usage

The package comes with a powerful CLI tool aid for encoding, decoding, and benchmarking.

   Arkadia AI DATA TOOL
   --------------------------------------------------
   Unified interface for AI Data Format operations.

USAGE:
   aid <command> [flags]

COMMANDS:
   enc             [ENCODE] Convert JSON/YAML/TOON to AI.Data format
   dec             [DECODE] Parse AI.Data format back to JSON
   benchmark       [BENCHMARK] Run performance and token usage tests
   ai-benchmark    [AI] Run AI understanding tests (not implemented yet)

GLOBAL OPTIONS:
   -h, --help       Show this help message
   -v, --version    Show version info

๐Ÿ“– Syntax Specification (Current Version)

This section describes the actual, currently implemented syntax of AI.DATA-FORMAT.

1. Type Definition

A type defines a name and an ordered list of fields. Comments are allowed within the definition to assist the LLM.

User</comment/ ={(23,"A",3) #tag1 #tag2} %[{ id: 4, b: "a", c: 43}]: id:number,
b: string , c:number, >
@Users
<
 @list 
 a: number,
 b: string
>
[
  @size=5
  /example list of values/

  (1,`text`,5)
  (2,`Text can be

multiline
`,5)
  {
    id:3,
    b: "text"
  }
]

Key Rules:

  • The type name (@Name) is optional but recommended.
  • The header <...> defines field names and their order.
  • Comments (/ ... /) are allowed in the header.

2. Data Structures

The format supports compact positional records and explicit named records.

Structure Syntax Description
Positional Record (a,b,c) Must follow the exact order of fields in the type header.
Named Record {key:value} Keys must match field names. No spaces allowed in keys/values.
List [ ... ] Contains positional or named records.
Multiline Text text Ends with a line containing only a backtick.

3. Comments

/ this is a comment /
  • Allowed only inside type definitions.
  • Forbidden in raw data blocks to save space.

4. General Rules

  1. Data must contain NO spaces. (Compactness is priority).
  2. Schema/Type definitions may contain spaces and comments.
  3. Named fields always use key:value without spaces.
  4. Positional order must exactly match the declared order.

5. Inline Type Usage

You can declare a type and immediately use it:

@User<id:number name:string desc:string>

value:@User(2,"Alice","Hello")
value2:@User(3,"Bob","World")

6. Nested Types

Currently, nested types are allowed as structural definitions:

@User<
  id:string
  name:string
  profile: < level:number, score:number >
>
[
  ("u1","Aga",{level:5,score:82})
  ("u2","Marek",{level:7,score:91})
]

๐Ÿ”ฎ Futures / Roadmap

The following features are planned for future releases and are not yet implemented.

  • Modifiers:

  • !required - field must be included.

  • ?empty - field must not be empty.

  • =value - default value.

  • N..M - numeric range validation.

  • Binary Data Types:

  • Hex: ~[hex]1A0F4F~

  • Base64: ~[b64]ADFKDXKZK...~

  • Pointers/References:

  • Reference existing objects by ID: (1, "Alex", *User[2])

๐Ÿ“„ License

This project is licensed under the [MIT License].


Built by Arkadia AI. Engineering the kernel of distributed intelligence.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

arkadia_ai_data_format-0.1.4.tar.gz (112.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

arkadia_ai_data_format-0.1.4-py3-none-any.whl (51.2 kB view details)

Uploaded Python 3

File details

Details for the file arkadia_ai_data_format-0.1.4.tar.gz.

File metadata

  • Download URL: arkadia_ai_data_format-0.1.4.tar.gz
  • Upload date:
  • Size: 112.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.10.0

File hashes

Hashes for arkadia_ai_data_format-0.1.4.tar.gz
Algorithm Hash digest
SHA256 1c2454a00d9d771ba1bdbf041021314f2e3838e72bef2acc5e835ed16689a41b
MD5 9d4b822dd9ef9a33ae3640e17742beae
BLAKE2b-256 e720f81739bc280a7243f8d0b9ed145d8802a5a65183d154d50048a9ce8a5daa

See more details on using hashes here.

File details

Details for the file arkadia_ai_data_format-0.1.4-py3-none-any.whl.

File metadata

File hashes

Hashes for arkadia_ai_data_format-0.1.4-py3-none-any.whl
Algorithm Hash digest
SHA256 5b55aaf524ddca699ece54cb4ef993180ccdea1227fcdd8bb03bf61924f08a34
MD5 c05ac8a9903758d3480c8428531ac80f
BLAKE2b-256 c195b77d9f2838648e5499b1871e7e10c9e01dd26b69bd8bba3b3ef2f25e2ad9

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page