Dataset Viewer
Auto-converted to Parquet Duplicate
instruction
stringclasses
4 values
messages
listlengths
2
2
label
stringclasses
2 values
မြန်မာ website တစ်ခုအတွက် Font ရွေးချယ်မှု Logic ကို ရှင်းပြပါ
[ { "role": "user", "content": "မြန်မာ website တစ်ခုအတွက် Font ရွေးချယ်မှု Logic ကို ရှင်းပြပါ" }, { "role": "assistant", "content": "### 🧠 AmkyawLogic: ၃ ဆင့်ယုတ္တိဗေဒ\n\n**၁။ Negation (ငြင်းပယ်ခြင်း)**\nမြန်မာ website များတွင် Zawgyi font ကို အဓိကထား၍ မသုံးစွဲရပါ။ ၎င်းသည် နိုင်ငံတကာ စံနှုန်း (U...
Web_UI_Logic
မြန်မာစာသားများအတွက် Line Height (စာကြောင်းအကွာအဝေး) Logic ကို ရှင်းပြပါ
[ { "role": "user", "content": "မြန်မာစာသားများအတွက် Line Height (စာကြောင်းအကွာအဝေး) Logic ကို ရှင်းပြပါ" }, { "role": "assistant", "content": "### 🧠 AmkyawLogic: ၃ ဆင့်ယုတ္တိဗေဒ\n\n**၁။ Negation (ငြင်းပယ်ခြင်း)**\nမြန်မာစာသားများအတွက် CSS line-height ကို အင်္ဂလိပ်စာကဲ့သို့ ကျဉ်းမြောင်းစွာ မထားရပ...
Web_UI_Logic
မြန်မာ website တစ်ခု၏ အရောင် (Color Palette) ရွေးချယ်မှု Logic ကို ရှင်းပြပါ
[ { "role": "user", "content": "မြန်မာ website တစ်ခု၏ အရောင် (Color Palette) ရွေးချယ်မှု Logic ကို ရှင်းပြပါ" }, { "role": "assistant", "content": "### 🧠 AmkyawLogic: ၃ ဆင့်ယုတ္တိဗေဒ\n\n**၁။ Negation (ငြင်းပယ်ခြင်း)**\nအရောင်ရွေးချယ်ရာတွင် ခေတ်သစ် (Modern) ဆန်ရုံတင်မကဘဲ မြန်မာ့ယဉ်ကျေးမှုနှင့် ဆန့...
Web_UI_Logic
မြန်မာ website တစ်ခု၏ Date and Time Format Logic ကို ရှင်းပြပါ
[ { "role": "user", "content": "မြန်မာ website တစ်ခု၏ Date and Time Format Logic ကို ရှင်းပြပါ" }, { "role": "assistant", "content": "### 🧠 AmkyawLogic: ၃ ဆင့်ယုတ္တိဗေဒ\n\n**၁။ Negation (ငြင်းပယ်ခြင်း)**\nနေ့စွဲများကို MM/DD/YYYY ပုံစံတစ်ခုတည်းဖြင့် မသတ်မှတ်ရပါ။ မြန်မာတို့သည် DD/MM/YYYY ကို ပိုမိ...
Web_UX_Logic

Amkyaw Core Ecosystem Dataset

အနှစ်ချုပ် (Overview)

Amkyaw Core Ecosystem သည် မြန်မာစကားအတွက် AI Reasoning Framework ဖြစ်ပါတယ်။

ဤDataset သည် Large Language Model (LLM) ကို "next-token prediction" မှ "identity-aware reasoning" သို့ ပြောင်းလဲရန် ရည်ရွယ်ပါတယ်။


Dataset Structure (4 Layers)

Layer 1: Identity Logic (အက္ခရာများ၏ သဘောသဘာဝ)

File Description
constants.json ဗျည်း၊ သရ၊ ဂဏန်းများ၏ Master List
alphabet_negation.jsonl "က သည် ခ မဟုတ်" စသည့် ငြင်းပယ်မှု logic
instruction_tuning.jsonl Instruction Tuning Template

Layer 2: Phonology (အသံထွက်နှင့် ပေါင်းစပ်ပုံ)

File Description
consonant_vowel_map.json ဗျည်းနှင့် သရ ပေါင်းစပ်မှု matrix
Layer_2_Phonology/syllable_logic.jsonl က + ာ = ကာ

Layer 3: Reasoning (AmkyawLogic - 3 Steps)

File Description
Layer_3_Reasoning/linguistic_reasoning.jsonl ဘာသာစကားဆိုင်ရာ ၃ ဆင့် တွေးခေါ်မှု
data/web_dev/myanmar_web_logic.jsonl Web UI/UX Logic

Layer 4: Gold Standard (အချောသတ်ဒေတာ)

File Description
Layer_4_Gold_Standard/train_ready.jsonl Final JSONL
data/consolidated_amkyaw_logic.jsonl အားလုံးကို ပေါင်းစပ်ထားသော

AmkyawLogic Framework (၃ ဆင့် ယုတ္တိဗေဒ)

The 3-Stage Reasoning Framework သည် ဤသို့ဖြစ်ပါတယ်။

Stage 1: Negation (ဆန့်ကျင်ဘက်)

  • မမှန်သည်များကို ဖယ်ထုတ်ခြင်း
  • Identity Boundary သတ်မှတ်ခြင်း

Stage 2: Structure (၅၀% မျှခြေ)

  • အလယ်အလတ်အမြင်
  • ပါးမိုးတည်ဆောက်ခြင်း

Stage 3: Optimized (အကောင်းဆုံး)

  • နောက်ဆုံး အဖြေထုတ်ပါတယ်။

JSONL Data Format

Format A: instruction_tuning.jsonl

{
  "instruction": "Explain the character [က] using AmkyawLogic.",
  "input": "က",
  "system_prompt": "You are a Reasoning Engine. Follow the 3-step synthesis: Negation, Structure, and Optimization.",
  "thought_trace": {
    "step_1": "Negation: Not 'ခ', not a number, not a symbol.",
    "step_2": "Structure: First consonant of the Burmese 33 characters.",
    "step_3": "Optimization: The foundational atom of the Burmese writing system."
  },
  "output": "['က'] is the primary consonant of the Burmese alphabet."
}

Format B: logic_steps (Web Dev, etc.)

{
  "instruction": "မြန်မာ website တစ်ခုအတွက် Font ရွေးချယ်မှု Logic ကို ရှင်းပြပါ",
  "label": "Web_UI_Logic",
  "logic_steps": {
    "step_1_negation": "မြန်မာ website များတွင် Zawgyi font ကို အဓိကထား၍ မသုံးစွဲရပါ။",
    "step_2_fifty_fifty": "Unicode font များဖြစ်သော Pyidaungsu သို့မဟုတ် Google Fonts မှ Padauk သုံးရမည်။",
    "step_3_optimized": "Standard Unicode ကို အခြေခံ၍ Responsive ဖြစ်သော font system ကို သုံးစွဲခြင်းဖြင့် မည်သည့် device တွင်မဆို မြန်မာစာသားများ မှန်ကန်လှပစွာ ပေါ်လွင်နေမည်။"
  }
}

Logic Steps Keys (JSON Keys)

Stage Key Description
1 (Negation) step_1_negation What it is NOT
2 (Structure/50-50) step_2_fifty_fifty Balanced perspective
3 (Optimized) step_3_optimized Best solution

Usage ပါ၊ (Python)

from datasets import load_dataset

# Load entire dataset
dataset = load_dataset("amkyawdev/amkyaw-core-ecosystem")

# Load specific JSONL file
web_logic = load_dataset(
    "amkyawdev/amkyaw-core-ecosystem", 
    data_files="data/web_dev/myanmar_web_logic.jsonl"
)

# Access logic_steps
for item in web_logic['train']:
    print(item['instruction'])
    print(item['logic_steps']['step_1_negation'])
    print(item['logic_steps']['step_2_fifty_fifty'])
    print(item['logic_steps']['step_3_optimized'])

ပါဝင်တဲ့ အရာများ (Categories)

  • Web UI/UX Logic: Font, Color, Line Height, Date Format
  • Linguistic Rules: Consonants, Vowels, Syllables
  • Daily Conversation: Question patterns, Negation
  • Numbers: Burmese numbers logic
  • Proverbs: Traditional proverbs
  • Geography: Myanmar geography identity

ပြင်ဆင်မှု လမ်းစဉ် (Implementation Roadmap)

  1. Tokenizer Calibration: Burmese Unicode/Zawgyi ကို မှန်ကန်ပါ၊ Character clusters မပိုကြာရပါ။

  2. Chain-of-Thought (CoT) Training: thought_trace field ကို သုံးပါ၊

  3. Cross-Domain Scaling: ၃ ဆင့် ယုတ္တိကို အခြား နယ်ပါလား (History, Science, Economy) မှာလည်း သုံးနိုင်ပါ။


License

MIT License

Author

Aung Myo Kyaw (amkyawdev)

Downloads last month
94

Space using amkyawdev/amkyaw-core-ecosystem 1