Update README.md - Add Technology Startup specialization and multi-AI provider support
Browse files
README.md
CHANGED
|
@@ -1,517 +1,517 @@
|
|
| 1 |
-
---
|
| 2 |
-
language:
|
| 3 |
-
- en
|
| 4 |
-
- my
|
| 5 |
-
license: apache-2.0
|
| 6 |
-
tags:
|
| 7 |
-
- business-intelligence
|
| 8 |
-
- sme
|
| 9 |
-
- myanmar
|
| 10 |
-
- diagnosis
|
| 11 |
-
- text-generation
|
| 12 |
-
- llama
|
| 13 |
-
- fine-tuned
|
| 14 |
-
- bios
|
| 15 |
-
- gold-shop
|
| 16 |
-
- southeast-asia
|
| 17 |
-
datasets:
|
| 18 |
-
- BIOS-kernel/myanmar-sme-diagnostics-v1
|
| 19 |
-
base_model: meta-llama/Llama-3.3-70B-Instruct
|
| 20 |
-
pipeline_tag: text-generation
|
| 21 |
-
model_type: causal-lm
|
| 22 |
-
widget:
|
| 23 |
-
- text: "Diagnose this business: Gold Shop in Yangon, 4.2M MMK monthly revenue, 28% retention rate, team of 3."
|
| 24 |
-
example_title: "Gold Shop Diagnosis"
|
| 25 |
-
- text: "What are the top growth opportunities for a Fashion business with 8M MMK revenue in Mandalay?"
|
| 26 |
-
example_title: "Fashion Growth Opportunities"
|
| 27 |
-
---
|
| 28 |
-
|
| 29 |
-
<div align="center">
|
| 30 |
-
|
| 31 |
-
```
|
| 32 |
-
╔══════════════════════════════════════════════════════════════╗
|
| 33 |
-
║ ║
|
| 34 |
-
║ ██████╗ ██╗ ██████╗ ███████╗ ║
|
| 35 |
-
║ ██╔══██╗██║██╔═══██╗██╔════╝ ║
|
| 36 |
-
║ ██████╔╝██║██║ ██║███████╗ ║
|
| 37 |
-
║ ██╔══██╗██║██║ ██║╚════██║ ║
|
| 38 |
-
║ ██████╔╝██║╚██████╔╝███████║ ║
|
| 39 |
-
║ ╚═════╝ ╚═╝ ╚═════╝ ╚══════╝ ║
|
| 40 |
-
║ ║
|
| 41 |
-
║ Business Idea Operating System ║
|
| 42 |
-
║ BIOS-Insight-v1 · Kernel: BIOS-kernel-v1 ║
|
| 43 |
-
║ ║
|
| 44 |
-
╚══════════════════════════════════════════════════════════════╝
|
| 45 |
-
```
|
| 46 |
-
|
| 47 |
-
**"We don't just analyse businesses. We illuminate them."**
|
| 48 |
-
|
| 49 |
-
[](LICENSE)
|
| 50 |
-
[](.)
|
| 51 |
-
[](.)
|
| 52 |
-
[](https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct)
|
| 53 |
-
[](.)
|
| 54 |
-
|
| 55 |
-
</div>
|
| 56 |
-
|
| 57 |
-
---
|
| 58 |
-
|
| 59 |
-
# BIOS-Insight-v1 — Business Idea Operating System
|
| 60 |
-
|
| 61 |
-
## 🇬🇧 English
|
| 62 |
-
|
| 63 |
-
### Model Description
|
| 64 |
-
|
| 65 |
-
**BIOS-Insight-v1** is a fine-tuned large language model built on **LLaMA 3.3 70B Instruct**, specifically trained to serve as the intelligence core of the **Business Idea Operating System (BIOS)** — a comprehensive AI agent designed for Myanmar's small and medium enterprises (SMEs), Gold Shops, fashion retailers, F&B operators, and the next generation of Southeast Asian entrepreneurs.
|
| 66 |
-
|
| 67 |
-
BIOS is not a chatbot. It is an **Operating System for business ideas** — the same way Windows runs your computer, BIOS runs your business strategy. Every question answered, every weakness surfaced, every opportunity ranked: all orchestrated by a single intelligent kernel.
|
| 68 |
-
|
| 69 |
-
This model powers **Module 1: Business Diagnosis Engine**, the foundational layer of the BIOS platform. Feed it 24 structured questions about any business, and it returns a complete, actionable diagnosis in under 60 seconds.
|
| 70 |
-
|
| 71 |
-
---
|
| 72 |
-
|
| 73 |
-
### Architecture & Training
|
| 74 |
-
|
| 75 |
-
| Property | Details |
|
| 76 |
-
|----------|---------|
|
| 77 |
-
| **Base Model** | `meta-llama/Llama-3.3-70B-Instruct` |
|
| 78 |
-
| **Fine-tune Method** | QLoRA (4-bit quantisation, rank 64) |
|
| 79 |
-
| **Training Data** | Myanmar SME diagnostics, Gold Shop patterns, SEA business benchmarks |
|
| 80 |
-
| **Context Length** | 8,192 tokens |
|
| 81 |
-
| **Output Format** | Structured JSON — deterministic, parseable |
|
| 82 |
-
| **Languages** | English, Burmese (မြန်မာဘာသာ) |
|
| 83 |
-
| **Quantisation** | GGUF Q4_K_M available for local inference |
|
| 84 |
-
|
| 85 |
-
---
|
| 86 |
-
|
| 87 |
-
### What BIOS Produces
|
| 88 |
-
|
| 89 |
-
Given structured business inputs, BIOS-Insight-v1 generates:
|
| 90 |
-
|
| 91 |
-
```json
|
| 92 |
-
{
|
| 93 |
-
"health_score": 47,
|
| 94 |
-
"health_label": "Fair",
|
| 95 |
-
"health_dimensions": {
|
| 96 |
-
"revenue_strength": 40,
|
| 97 |
-
"customer_retention": 20,
|
| 98 |
-
"market_position": 60,
|
| 99 |
-
"technology_adoption": 30,
|
| 100 |
-
"growth_trajectory": 80
|
| 101 |
-
},
|
| 102 |
-
"top_3_weaknesses": [
|
| 103 |
-
{
|
| 104 |
-
"rank": 1,
|
| 105 |
-
"label": "Customer Retention",
|
| 106 |
-
"your_score": 20,
|
| 107 |
-
"benchmark": 60,
|
| 108 |
-
"gap": 40,
|
| 109 |
-
"severity": "HIGH",
|
| 110 |
-
"detail": "Only 28% repeat purchase rate — Gold Shop industry average is 60%."
|
| 111 |
-
}
|
| 112 |
-
],
|
| 113 |
-
"growth_opportunities": [
|
| 114 |
-
{
|
| 115 |
-
"rank": 1,
|
| 116 |
-
"title": "Boost Customer Retention Rate",
|
| 117 |
-
"expected_impact": "+1,680,000 MMK estimated monthly revenue",
|
| 118 |
-
"difficulty": "MEDIUM",
|
| 119 |
-
"timeframe": "2–3 months"
|
| 120 |
-
}
|
| 121 |
-
],
|
| 122 |
-
"priority_action_items": [
|
| 123 |
-
{
|
| 124 |
-
"priority": 1,
|
| 125 |
-
"action": "Launch a loyalty stamp card and 30-day WhatsApp follow-up sequence.",
|
| 126 |
-
"composite_score": 82.0
|
| 127 |
-
}
|
| 128 |
-
],
|
| 129 |
-
"ai_narrative": "Shwe Zin Gold & Jewellery is operating at 47/100 health — a Fair rating that conceals a serious retention gap..."
|
| 130 |
-
}
|
| 131 |
-
```
|
| 132 |
-
|
| 133 |
-
---
|
| 134 |
-
|
| 135 |
-
### Health Score Formula
|
| 136 |
-
|
| 137 |
-
The BIOS Health Score is calculated across five equally-weighted dimensions:
|
| 138 |
-
|
| 139 |
-
```
|
| 140 |
-
Health Score = (Revenue Strength × 20%) +
|
| 141 |
-
(Customer Retention × 20%) +
|
| 142 |
-
(Market Position × 20%) +
|
| 143 |
-
(Technology Adoption × 20%) +
|
| 144 |
-
(Growth Trajectory × 20%)
|
| 145 |
-
|
| 146 |
-
Where each dimension is scored 0–100.
|
| 147 |
-
Maximum Score: 100
|
| 148 |
-
```
|
| 149 |
-
|
| 150 |
-
| Score Range | Label | Interpretation |
|
| 151 |
-
|------------|-------|---------------|
|
| 152 |
-
| 80 – 100 | 🟢 Excellent | Best-in-class. Scale aggressively. |
|
| 153 |
-
| 65 – 79 | 🔵 Good | Strong foundation. Focus on 1–2 gaps. |
|
| 154 |
-
| 45 – 64 | 🟡 Fair | Visible weaknesses. Targeted fixes needed. |
|
| 155 |
-
| 30 – 44 | 🟠 Below Average | Systemic issues. Restructure required. |
|
| 156 |
-
| 0 – 29 | 🔴 Critical | Immediate intervention. Prioritise survival. |
|
| 157 |
-
|
| 158 |
-
---
|
| 159 |
-
|
| 160 |
-
### Intended Use
|
| 161 |
-
|
| 162 |
-
#### ✅ Primary Use Cases
|
| 163 |
-
|
| 164 |
-
- **Myanmar Gold Shops & Jewellers** — the lifeblood of Myanmar's retail economy, underserved by digital tools
|
| 165 |
-
- **Fashion & Apparel SMEs** — fast-moving businesses in Yangon, Mandalay, Naypyidaw
|
| 166 |
-
- **F&B Operators** — restaurants, tea shops, catering businesses
|
| 167 |
-
- **Cosmetics & Beauty Brands** — direct-to-consumer Myanmar brands scaling up
|
| 168 |
-
- **Electronics Retailers** — high-value, low-margin businesses needing operational precision
|
| 169 |
-
- **Any Myanmar SME founder** who wants strategic clarity without a consultant's fee
|
| 170 |
-
|
| 171 |
-
#### ❌ Out-of-Scope Uses
|
| 172 |
-
|
| 173 |
-
- Large corporations (BIOS is tuned for SME scale and context)
|
| 174 |
-
- Non-business tasks (general Q&A, creative writing)
|
| 175 |
-
- Legal or financial advice (BIOS provides business intelligence, not regulated advisory)
|
| 176 |
-
|
| 177 |
-
---
|
| 178 |
-
|
| 179 |
-
### How to Use
|
| 180 |
-
|
| 181 |
-
#### With the `transformers` Library
|
| 182 |
-
|
| 183 |
-
```python
|
| 184 |
-
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 185 |
-
import torch
|
| 186 |
-
|
| 187 |
-
model_id = "BIOS-kernel/BIOS-Insight-v1"
|
| 188 |
-
|
| 189 |
-
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 190 |
-
model = AutoModelForCausalLM.from_pretrained(
|
| 191 |
-
model_id,
|
| 192 |
-
torch_dtype=torch.bfloat16,
|
| 193 |
-
device_map="auto",
|
| 194 |
-
)
|
| 195 |
-
|
| 196 |
-
system_prompt = """You are BIOS — the Business Idea Operating System.
|
| 197 |
-
You are the elite AI advisor for Myanmar SMEs.
|
| 198 |
-
Always respond in valid JSON with health_score, top_3_weaknesses,
|
| 199 |
-
growth_opportunities, and priority_action_items."""
|
| 200 |
-
|
| 201 |
-
user_prompt = """Diagnose this business:
|
| 202 |
-
Business: Shwe Zin Gold & Jewellery | Industry: Gold Shop | Location: Yangon
|
| 203 |
-
Monthly Revenue: 4,200,000 MMK | Retention Rate: 28% | Team: 3 people
|
| 204 |
-
USP: Certified 99.9% pure gold with 10-year buyback guarantee
|
| 205 |
-
Pain Point: No customer follow-up system. Customers don't return.
|
| 206 |
-
12-Month Goal: 12,000,000 MMK
|
| 207 |
-
Marketing Budget: 80,000 MMK/month"""
|
| 208 |
-
|
| 209 |
-
messages = [
|
| 210 |
-
{"role": "system", "content": system_prompt},
|
| 211 |
-
{"role": "user", "content": user_prompt},
|
| 212 |
-
]
|
| 213 |
-
|
| 214 |
-
input_ids = tokenizer.apply_chat_template(
|
| 215 |
-
messages,
|
| 216 |
-
add_generation_prompt=True,
|
| 217 |
-
return_tensors="pt",
|
| 218 |
-
).to(model.device)
|
| 219 |
-
|
| 220 |
-
output = model.generate(
|
| 221 |
-
input_ids,
|
| 222 |
-
max_new_tokens=1024,
|
| 223 |
-
temperature=0.3,
|
| 224 |
-
do_sample=True,
|
| 225 |
-
pad_token_id=tokenizer.eos_token_id,
|
| 226 |
-
)
|
| 227 |
-
|
| 228 |
-
response = tokenizer.decode(
|
| 229 |
-
output[0][input_ids.shape[-1]:],
|
| 230 |
-
skip_special_tokens=True,
|
| 231 |
-
)
|
| 232 |
-
print(response)
|
| 233 |
-
```
|
| 234 |
-
|
| 235 |
-
#### With the BIOS Controller (Recommended)
|
| 236 |
-
|
| 237 |
-
```python
|
| 238 |
-
from bios_controller import BIOSController, BusinessInputs, ModelBackend
|
| 239 |
-
|
| 240 |
-
# Initialise
|
| 241 |
-
controller = BIOSController(
|
| 242 |
-
backend = ModelBackend.GROQ, # or HF_INFERENCE when BIOS-Insight-v1 is live
|
| 243 |
-
save_to_db = True, # persist to NeonDB
|
| 244 |
-
)
|
| 245 |
-
|
| 246 |
-
# Fill in the 24 business questions
|
| 247 |
-
inputs = BusinessInputs(
|
| 248 |
-
business_name = "Shwe Zin Gold & Jewellery",
|
| 249 |
-
industry = "Gold Shop",
|
| 250 |
-
location = "Yangon",
|
| 251 |
-
years_in_business = 7,
|
| 252 |
-
monthly_revenue = 4_200_000,
|
| 253 |
-
team_size = 3,
|
| 254 |
-
target_customer = "Middle-income families, 30–55, buying gold for investment and festivals",
|
| 255 |
-
acquisition_channels = ["Word-of-mouth", "Facebook", "Walk-in"],
|
| 256 |
-
avg_customer_lifetime_value= 350_000,
|
| 257 |
-
retention_rate = 28.0,
|
| 258 |
-
main_competitors = "Dagon Gold, KBZ Gems",
|
| 259 |
-
unique_selling_proposition = "Certified 99.9% gold. Transparent pricing. 10-year buyback guarantee.",
|
| 260 |
-
sales_channels = ["Physical Store", "Facebook"],
|
| 261 |
-
operational_challenge = "Inventory management",
|
| 262 |
-
biggest_pain_point = "No system to follow up with customers after first purchase.",
|
| 263 |
-
current_technology = ["Spreadsheets"],
|
| 264 |
-
marketing_channels = ["Facebook", "Word-of-mouth"],
|
| 265 |
-
monthly_marketing_budget = 80_000,
|
| 266 |
-
goal_3_month = 5_500_000,
|
| 267 |
-
goal_6_month = 7_000_000,
|
| 268 |
-
goal_12_month = 12_000_000,
|
| 269 |
-
budget_constraint = "Tight (50-200K)",
|
| 270 |
-
tech_readiness = "Somewhat ready",
|
| 271 |
-
preferred_language = "English",
|
| 272 |
-
)
|
| 273 |
-
|
| 274 |
-
# Run the full diagnosis pipeline
|
| 275 |
-
report = controller.run_diagnosis(inputs)
|
| 276 |
-
|
| 277 |
-
# Access structured results
|
| 278 |
-
print(f"Health Score : {report.health_score}/100 ({report.health_label})")
|
| 279 |
-
print(f"Top Weakness : {report.top_3_weaknesses[0].label}")
|
| 280 |
-
print(f"Best Opportunity : {report.growth_opportunities[0].title}")
|
| 281 |
-
print(f"\nAI Narrative:\n{report.ai_narrative}")
|
| 282 |
-
|
| 283 |
-
# Full JSON output
|
| 284 |
-
print(report.to_json())
|
| 285 |
-
```
|
| 286 |
-
|
| 287 |
-
#### With HuggingFace Inference API
|
| 288 |
-
|
| 289 |
-
```python
|
| 290 |
-
from huggingface_hub import InferenceClient
|
| 291 |
-
|
| 292 |
-
client = InferenceClient(
|
| 293 |
-
model = "BIOS-kernel/BIOS-Insight-v1",
|
| 294 |
-
token = "hf_your_token_here",
|
| 295 |
-
)
|
| 296 |
-
|
| 297 |
-
response = client.chat_completion(
|
| 298 |
-
messages=[
|
| 299 |
-
{"role": "system", "content": "You are BIOS. Respond in JSON."},
|
| 300 |
-
{"role": "user", "content": "Diagnose: Gold Shop, 4.2M MMK revenue, 28% retention."},
|
| 301 |
-
],
|
| 302 |
-
max_tokens = 1024,
|
| 303 |
-
temperature = 0.3,
|
| 304 |
-
)
|
| 305 |
-
print(response.choices[0].message.content)
|
| 306 |
-
```
|
| 307 |
-
|
| 308 |
-
---
|
| 309 |
-
|
| 310 |
-
### Switching Models (Base vs Fine-tuned)
|
| 311 |
-
|
| 312 |
-
```python
|
| 313 |
-
controller = BIOSController(backend=ModelBackend.GROQ)
|
| 314 |
-
|
| 315 |
-
# Use base LLaMA-3.3-70B (default, available now)
|
| 316 |
-
report_base = controller.run_diagnosis(inputs)
|
| 317 |
-
|
| 318 |
-
# Switch to BIOS-Insight-v1 once published on HuggingFace
|
| 319 |
-
controller.switch_to_bios_insight()
|
| 320 |
-
report_bios = controller.run_diagnosis(inputs)
|
| 321 |
-
|
| 322 |
-
# Switch back to base
|
| 323 |
-
controller.switch_to_base()
|
| 324 |
-
```
|
| 325 |
-
|
| 326 |
-
---
|
| 327 |
-
|
| 328 |
-
### NeonDB Integration
|
| 329 |
-
|
| 330 |
-
```python
|
| 331 |
-
import os
|
| 332 |
-
os.environ["DATABASE_URL"] = "postgresql://user:pass@ep-xxx.neon.tech/neondb?sslmode=require"
|
| 333 |
-
|
| 334 |
-
controller = BIOSController(save_to_db=True)
|
| 335 |
-
report = controller.run_diagnosis(inputs)
|
| 336 |
-
|
| 337 |
-
# Retrieve saved report
|
| 338 |
-
saved = controller.get_report(report.session_id)
|
| 339 |
-
|
| 340 |
-
# List all diagnoses
|
| 341 |
-
history = controller.list_reports(limit=10)
|
| 342 |
-
```
|
| 343 |
-
|
| 344 |
-
---
|
| 345 |
-
|
| 346 |
-
### Limitations
|
| 347 |
-
|
| 348 |
-
- Benchmarks are calibrated for Myanmar market (MMK currency, Yangon/Mandalay/Naypyidaw context)
|
| 349 |
-
- Growth projections are estimates, not guarantees — market conditions vary
|
| 350 |
-
- The model does not access real-time data or the internet
|
| 351 |
-
- Legal and financial decisions should always be reviewed by qualified professionals
|
| 352 |
-
|
| 353 |
-
---
|
| 354 |
-
|
| 355 |
-
### Citation
|
| 356 |
-
|
| 357 |
-
```bibtex
|
| 358 |
-
@misc{bios-insight-v1,
|
| 359 |
-
title = {BIOS-Insight-v1: Business Idea Operating System for Myanmar SMEs},
|
| 360 |
-
author = {BIOS-kernel},
|
| 361 |
-
year = {2026},
|
| 362 |
-
howpublished = {\url{https://huggingface.co/BIOS-kernel/BIOS-Insight-v1}},
|
| 363 |
-
note = {Fine-tuned on LLaMA 3.3 70B Instruct for Myanmar business diagnostics}
|
| 364 |
-
}
|
| 365 |
-
```
|
| 366 |
-
|
| 367 |
-
---
|
| 368 |
-
|
| 369 |
-
---
|
| 370 |
-
|
| 371 |
-
## 🇲🇲 မြန်မာဘာသာ (Burmese)
|
| 372 |
-
|
| 373 |
-
### မော်ဒယ်ဖော်ပြချက်
|
| 374 |
-
|
| 375 |
-
**BIOS-Insight-v1** သည် **LLaMA 3.3 70B Instruct** ကို အခြေခံ၍ fine-tune ပြုလုပ်ထားသော AI မော်ဒယ်တစ်ခုဖြစ်ပြီး၊ မြန်မာနိုင်ငံ၏ SME (အသေးစားနှင့် အလတ်စားလုပ်ငန်းများ) — ရွှေဆိုင်များ၊ ဖက်ရှင်ဆိုင်များ၊ စားသောက်ဆိုင်များ၊ နှင့် နောင်လာမည့် Southeast Asia ၏ လုပ်ငန်းရှင်များအတွက် **Business Idea Operating System (BIOS)** ၏ AI အဓိကအင်ဂျင်အဖြစ် ဒီဇိုင်းထုတ်ထားသည်။
|
| 376 |
-
|
| 377 |
-
BIOS သည် chatbot တစ်ခုမဟုတ်ပါ။ ၎င်းသည် **သင်၏လုပ်ငန်းအကြံဥာဏ်များအတွက် Operating System** တစ်ခုဖြစ်သည် — Windows က သင်၏ကွန်ပျူတာကို run သကဲ့သို့၊ BIOS က သင်၏လုပ်ငန်းဗျူဟာကို run သည်။ မေးထားသောမေးခွန်းတိုင်း၊ ဖော်ထုတ်သော အားနည်းချက်တိုင်း၊ အဆင့်သတ်မှတ်ထားသော အခွင့်အလမ်းတိုင်း — ဆောင်ရွက်မှုအားလုံးကို AI kernel တစ်ခုတည်းဖြင့် လမ်းညွှန်သည်။
|
| 378 |
-
|
| 379 |
-
---
|
| 380 |
-
|
| 381 |
-
### ရည်ရွယ်သောအသုံးပြုနယ်ပယ်
|
| 382 |
-
|
| 383 |
-
BIOS-Insight-v1 ကို အောက်ပါလုပ်ငန်းများအတွက် အထူးသင့်တော်သည်:
|
| 384 |
-
|
| 385 |
-
**✅ အဓိကအသုံးပြုနယ်ပယ်များ**
|
| 386 |
-
|
| 387 |
-
- 🥇 **မြန်မာရွှေဆိုင်များနှင့် လက်ဝတ်ရတနာဆိုင်များ** — မြန်မာ့လက်လီကုန်ခြောက်စီးပွားရေး၏ အသက်ကြောဖြစ်သော ဆိုင်များ
|
| 388 |
-
- 👗 **ဖက်ရှင်နှင့် အဝတ်အထည် SME များ** — ရန်ကုန်၊ မန္တလေး၊ နေပြည်တော်ရှိ ဆိုင်များ
|
| 389 |
-
- 🍜 **F&B လုပ်ငန်းများ** — စားသောက်ဆိုင်၊ လက်ဖက်ရည်ဆိုင်၊ Catering လုပ်ငန်းများ
|
| 390 |
-
- 💄 **လှပရေးနှင့် ကောင်မီတစ်ဆ Brand များ** — မြန်မာ DTC Brand များ
|
| 391 |
-
- 📱 **Electronics ဆိုင်များ** — ကုန်ပစ္စည်းတန်ဖိုးမြင့်သော၊ margin နည်းသောလုပ်ငန်းများ
|
| 392 |
-
- 🏢 **မြန်မာ SME တည်ထောင်သူများ** — consultant ဦးစောင်ကြေးမပေးဘဲ ဗျူဟာကို ရှင်းလင်းစေလိုသူများ
|
| 393 |
-
|
| 394 |
-
---
|
| 395 |
-
|
| 396 |
-
### BIOS ၏ ကျန်းမာရေးရမှတ်ဖော်မြူလာ
|
| 397 |
-
|
| 398 |
-
BIOS Health Score ကို ညီမျှသောအချိန်ချိန်ထားသော ကဏ္ဍ ၅ ခုဖြင့် တွက်ချက်သည်:
|
| 399 |
-
|
| 400 |
-
```
|
| 401 |
-
Health Score = (ဝင်ငွေခိုင်ခံ့မှု × ၂၀%) +
|
| 402 |
-
(ဖောက်သည်ဆက်လက်ဆောင်ရွက်မှု × ၂၀%) +
|
| 403 |
-
(ဈေးကွက်တွင်နေရာ × ၂၀%) +
|
| 404 |
-
(နည်းပညာဆိုင်ရာသုံးစွဲမှု × ၂၀%) +
|
| 405 |
-
(တိုးတက်မှ
|
| 406 |
-
|
| 407 |
-
အမြင့်ဆုံးရမှတ်: ၁၀၀
|
| 408 |
-
```
|
| 409 |
-
|
| 410 |
-
| ရမှတ် | အမှတ်တံဆိပ် | အဓိပ္ပါယ် |
|
| 411 |
-
|------|------------|---------|
|
| 412 |
-
| ၈၀–၁၀၀ | 🟢 ထူးခြားကောင်းမွန်သော | ကဏ္ဍ အကောင်းဆုံး။ တိုးချဲ့ပါ။ |
|
| 413 |
-
| ၆၅–၇၉ | 🔵 ကောင်းမွန်သော | ခိုင်မာသောအခြေခံ။ ကွာဟချက် ၁–၂ ခုကို အာရုံစိုက်ပါ။ |
|
| 414 |
-
| ၄၅–၆၄ | 🟡 ဖြစ်နိုင်သော | မြင်သာသောအားနည်းချက်များ။ ပစ်မှတ်ထားပြင်ဆင်ရန်လိုသည်။ |
|
| 415 |
-
| ၃၀–၄၄ | 🟠 ပျမ်းမျှအောက် | စနစ်ဆိုင်ရာပြဿနာများ။ ပြန်ဖွဲ့စည်းရန်လိုသည်။ |
|
| 416 |
-
| ၀–၂၉ | 🔴 အရေးပေါ် | ချက်ချင်းဝင်ရောက်ကူညီရန်လိုသည်။ |
|
| 417 |
-
|
| 418 |
-
---
|
| 419 |
-
|
| 420 |
-
### မည်သို့အသုံးပြုမည်နည်း (`transformers` နှင့်)
|
| 421 |
-
|
| 422 |
-
```python
|
| 423 |
-
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 424 |
-
import torch
|
| 425 |
-
|
| 426 |
-
model_id = "BIOS-kernel/BIOS-Insight-v1"
|
| 427 |
-
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 428 |
-
model = AutoModelForCausalLM.from_pretrained(
|
| 429 |
-
model_id,
|
| 430 |
-
torch_dtype = torch.bfloat16,
|
| 431 |
-
device_map = "auto",
|
| 432 |
-
)
|
| 433 |
-
|
| 434 |
-
# မြန်မာဘာသာဖြင့် မေးမြန်းနိုင်သည်
|
| 435 |
-
messages = [
|
| 436 |
-
{
|
| 437 |
-
"role": "system",
|
| 438 |
-
"content": (
|
| 439 |
-
"သင်သည် BIOS ဖြစ်သည် — Business Idea Operating System။ "
|
| 440 |
-
"မြန်မာ SME များအတွက် elite AI အကြံပေး။ "
|
| 441 |
-
"JSON ဖော်မတ်ဖြင့် ဖြေပါ။"
|
| 442 |
-
),
|
| 443 |
-
},
|
| 444 |
-
{
|
| 445 |
-
"role": "user",
|
| 446 |
-
"content": (
|
| 447 |
-
"ဤလုပ်ငန်းကို စစ်ဆေးပါ:\n"
|
| 448 |
-
"လုပ်ငန်း: ရွှေဇင် ရွှေနှင့် လက်ဝတ်ရတနာ | ကဏ္ဍ: ရွှေဆိုင် | တည်နေရာ: ရန်ကုန်\n"
|
| 449 |
-
"လစဉ်ဝင်ငွေ: ၄,၂၀၀,၀၀၀ ကျပ် | Retention Rate: ၂၈% | အဖွဲ့ဝင်: ၃ ဦး\n"
|
| 450 |
-
"အကြီးဆုံးပြဿနာ: ဖောက်သည်များကို ပြန်မလာအောင် ဆက်သွယ်နိုင်သောစနစ် မရှိ\n"
|
| 451 |
-
"၁၂ လပန်းတိုင်: ၁၂,၀၀၀,၀၀၀ ကျပ်"
|
| 452 |
-
),
|
| 453 |
-
},
|
| 454 |
-
]
|
| 455 |
-
|
| 456 |
-
input_ids = tokenizer.apply_chat_template(
|
| 457 |
-
messages, add_generation_prompt=True, return_tensors="pt"
|
| 458 |
-
).to(model.device)
|
| 459 |
-
|
| 460 |
-
output = model.generate(
|
| 461 |
-
input_ids, max_new_tokens=1024, temperature=0.3, do_sample=True,
|
| 462 |
-
pad_token_id=tokenizer.eos_token_id,
|
| 463 |
-
)
|
| 464 |
-
response = tokenizer.decode(output[0][input_ids.shape[-1]:], skip_special_tokens=True)
|
| 465 |
-
print(response)
|
| 466 |
-
```
|
| 467 |
-
|
| 468 |
-
---
|
| 469 |
-
|
| 470 |
-
### BIOS Controller ဖြင့် အသုံးပြုခြင်း
|
| 471 |
-
|
| 472 |
-
```python
|
| 473 |
-
from bios_controller import BIOSController, BusinessInputs, ModelBackend
|
| 474 |
-
|
| 475 |
-
controller = BIOSController(backend=ModelBackend.GROQ, save_to_db=True)
|
| 476 |
-
|
| 477 |
-
inputs = BusinessInputs(
|
| 478 |
-
business_name = "ရွှေဇင် ရွှေနှင့် လက်ဝတ်ရတနာ",
|
| 479 |
-
industry = "Gold Shop",
|
| 480 |
-
location = "ရန်ကုန်",
|
| 481 |
-
years_in_business = 7,
|
| 482 |
-
monthly_revenue = 4_200_000,
|
| 483 |
-
team_size = 3,
|
| 484 |
-
retention_rate = 28.0,
|
| 485 |
-
unique_selling_proposition = "အသိအမှတ်ပြုထားသော ၉၉.၉% ရွှေစစ် — ၁၀ နှစ် buyback အာမခံ",
|
| 486 |
-
biggest_pain_point = "ဖောက်သည်များကို ပထမဝယ်ပြီးနောက် ဆက်သွယ်နိုင်သောစနစ် မရှိ",
|
| 487 |
-
goal_12_month = 12_000_000,
|
| 488 |
-
preferred_language = "မြန်မာဘာသာ",
|
| 489 |
-
# ... (မေးခွန်း ၂၄ ခုလုံး)
|
| 490 |
-
)
|
| 491 |
-
|
| 492 |
-
report = controller.run_diagnosis(inputs)
|
| 493 |
-
print(f"ကျန်းမာရေးရမှတ်: {report.health_score}/၁၀၀ ({report.health_label})")
|
| 494 |
-
print(f"AI အစီရင်ခံချက်:\n{report.ai_narrative}")
|
| 495 |
-
```
|
| 496 |
-
|
| 497 |
-
---
|
| 498 |
-
|
| 499 |
-
### လုံခြုံရေးနှင့် ကန့်သတ်ချက်များ
|
| 500 |
-
|
| 501 |
-
- Benchmark များသည် မြန်မာ့ဈေးကွက်အခြေအနေ (MMK ငွေကြေး) အတွက် ချိန်ညှိထားသည်
|
| 502 |
-
- ကြီးထွားမှုခန့်မှန်းချက်များသည် estimate များသာဖြစ်ပြီး အာမခံချက်မပေးနိုင်ပါ
|
| 503 |
-
- ဥပဒေနှင့် ဘဏ္ဍာရေးဆိုင်ရာ ဆုံးဖြတ်ချက်များကို အရည်အချင်းပြည့်ဝသောကျွမ်းကျင်သူများနှင့် ပြန်လည်စစ်ဆေးသင့်သည်
|
| 504 |
-
|
| 505 |
-
---
|
| 506 |
-
|
| 507 |
-
<div align="center">
|
| 508 |
-
|
| 509 |
-
**BIOS — Business Idea Operating System**
|
| 510 |
-
|
| 511 |
-
*"သင်၏လုပ်ငန်းကို ကျွန်ုပ်တို့ ရိုးရိုးစစ်ဆေးတာမဟုတ်ပါ။ ကျွန်ုပ်တို့ ၎င်းကို လင်းထိန်စေသည်။"*
|
| 512 |
-
|
| 513 |
-
*"We don't just analyse businesses. We illuminate them."*
|
| 514 |
-
|
| 515 |
-
[](https://huggingface.co/BIOS-kernel/BIOS-Insight-v1)
|
| 516 |
-
|
| 517 |
-
</div>
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
+
- my
|
| 5 |
+
license: apache-2.0
|
| 6 |
+
tags:
|
| 7 |
+
- business-intelligence
|
| 8 |
+
- sme
|
| 9 |
+
- myanmar
|
| 10 |
+
- diagnosis
|
| 11 |
+
- text-generation
|
| 12 |
+
- llama
|
| 13 |
+
- fine-tuned
|
| 14 |
+
- bios
|
| 15 |
+
- gold-shop
|
| 16 |
+
- southeast-asia
|
| 17 |
+
datasets:
|
| 18 |
+
- BIOS-kernel/myanmar-sme-diagnostics-v1
|
| 19 |
+
base_model: meta-llama/Llama-3.3-70B-Instruct
|
| 20 |
+
pipeline_tag: text-generation
|
| 21 |
+
model_type: causal-lm
|
| 22 |
+
widget:
|
| 23 |
+
- text: "Diagnose this business: Gold Shop in Yangon, 4.2M MMK monthly revenue, 28% retention rate, team of 3."
|
| 24 |
+
example_title: "Gold Shop Diagnosis"
|
| 25 |
+
- text: "What are the top growth opportunities for a Fashion business with 8M MMK revenue in Mandalay?"
|
| 26 |
+
example_title: "Fashion Growth Opportunities"
|
| 27 |
+
---
|
| 28 |
+
|
| 29 |
+
<div align="center">
|
| 30 |
+
|
| 31 |
+
```
|
| 32 |
+
╔══════════════════════════════════════════════════════════════╗
|
| 33 |
+
║ ║
|
| 34 |
+
║ ██████╗ ██╗ ██████╗ ███████╗ ║
|
| 35 |
+
║ ██╔══██╗██║██╔═══██╗██╔════╝ ║
|
| 36 |
+
║ ██████╔╝██║██║ ██║███████╗ ║
|
| 37 |
+
║ ██╔══██╗██║██║ ██║╚════██║ ║
|
| 38 |
+
║ ██████╔╝██║╚██████╔╝███████║ ║
|
| 39 |
+
║ ╚═════╝ ╚═╝ ╚═════╝ ╚══════╝ ║
|
| 40 |
+
║ ║
|
| 41 |
+
║ Business Idea Operating System ║
|
| 42 |
+
║ BIOS-Insight-v1 · Kernel: BIOS-kernel-v1 ║
|
| 43 |
+
║ ║
|
| 44 |
+
╚══════════════════════════════════════════════════════════════╝
|
| 45 |
+
```
|
| 46 |
+
|
| 47 |
+
**"We don't just analyse businesses. We illuminate them."**
|
| 48 |
+
|
| 49 |
+
[](LICENSE)
|
| 50 |
+
[](.)
|
| 51 |
+
[](.)
|
| 52 |
+
[](https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct)
|
| 53 |
+
[](.)
|
| 54 |
+
|
| 55 |
+
</div>
|
| 56 |
+
|
| 57 |
+
---
|
| 58 |
+
|
| 59 |
+
# BIOS-Insight-v1 — Business Idea Operating System
|
| 60 |
+
|
| 61 |
+
## 🇬🇧 English
|
| 62 |
+
|
| 63 |
+
### Model Description
|
| 64 |
+
|
| 65 |
+
**BIOS-Insight-v1** is a fine-tuned large language model built on **LLaMA 3.3 70B Instruct**, specifically trained to serve as the intelligence core of the **Business Idea Operating System (BIOS)** — a comprehensive AI agent designed for Myanmar's small and medium enterprises (SMEs), Gold Shops, fashion retailers, F&B operators, and the next generation of Southeast Asian entrepreneurs.
|
| 66 |
+
|
| 67 |
+
BIOS is not a chatbot. It is an **Operating System for business ideas** — the same way Windows runs your computer, BIOS runs your business strategy. Every question answered, every weakness surfaced, every opportunity ranked: all orchestrated by a single intelligent kernel.
|
| 68 |
+
|
| 69 |
+
This model powers **Module 1: Business Diagnosis Engine**, the foundational layer of the BIOS platform. Feed it 24 structured questions about any business, and it returns a complete, actionable diagnosis in under 60 seconds.
|
| 70 |
+
|
| 71 |
+
---
|
| 72 |
+
|
| 73 |
+
### Architecture & Training
|
| 74 |
+
|
| 75 |
+
| Property | Details |
|
| 76 |
+
|----------|---------|
|
| 77 |
+
| **Base Model** | `meta-llama/Llama-3.3-70B-Instruct` |
|
| 78 |
+
| **Fine-tune Method** | QLoRA (4-bit quantisation, rank 64) |
|
| 79 |
+
| **Training Data** | Myanmar SME diagnostics, Gold Shop patterns, SEA business benchmarks |
|
| 80 |
+
| **Context Length** | 8,192 tokens |
|
| 81 |
+
| **Output Format** | Structured JSON — deterministic, parseable |
|
| 82 |
+
| **Languages** | English, Burmese (မြန်မာဘာသာ) |
|
| 83 |
+
| **Quantisation** | GGUF Q4_K_M available for local inference |
|
| 84 |
+
|
| 85 |
+
---
|
| 86 |
+
|
| 87 |
+
### What BIOS Produces
|
| 88 |
+
|
| 89 |
+
Given structured business inputs, BIOS-Insight-v1 generates:
|
| 90 |
+
|
| 91 |
+
```json
|
| 92 |
+
{
|
| 93 |
+
"health_score": 47,
|
| 94 |
+
"health_label": "Fair",
|
| 95 |
+
"health_dimensions": {
|
| 96 |
+
"revenue_strength": 40,
|
| 97 |
+
"customer_retention": 20,
|
| 98 |
+
"market_position": 60,
|
| 99 |
+
"technology_adoption": 30,
|
| 100 |
+
"growth_trajectory": 80
|
| 101 |
+
},
|
| 102 |
+
"top_3_weaknesses": [
|
| 103 |
+
{
|
| 104 |
+
"rank": 1,
|
| 105 |
+
"label": "Customer Retention",
|
| 106 |
+
"your_score": 20,
|
| 107 |
+
"benchmark": 60,
|
| 108 |
+
"gap": 40,
|
| 109 |
+
"severity": "HIGH",
|
| 110 |
+
"detail": "Only 28% repeat purchase rate — Gold Shop industry average is 60%."
|
| 111 |
+
}
|
| 112 |
+
],
|
| 113 |
+
"growth_opportunities": [
|
| 114 |
+
{
|
| 115 |
+
"rank": 1,
|
| 116 |
+
"title": "Boost Customer Retention Rate",
|
| 117 |
+
"expected_impact": "+1,680,000 MMK estimated monthly revenue",
|
| 118 |
+
"difficulty": "MEDIUM",
|
| 119 |
+
"timeframe": "2–3 months"
|
| 120 |
+
}
|
| 121 |
+
],
|
| 122 |
+
"priority_action_items": [
|
| 123 |
+
{
|
| 124 |
+
"priority": 1,
|
| 125 |
+
"action": "Launch a loyalty stamp card and 30-day WhatsApp follow-up sequence.",
|
| 126 |
+
"composite_score": 82.0
|
| 127 |
+
}
|
| 128 |
+
],
|
| 129 |
+
"ai_narrative": "Shwe Zin Gold & Jewellery is operating at 47/100 health — a Fair rating that conceals a serious retention gap..."
|
| 130 |
+
}
|
| 131 |
+
```
|
| 132 |
+
|
| 133 |
+
---
|
| 134 |
+
|
| 135 |
+
### Health Score Formula
|
| 136 |
+
|
| 137 |
+
The BIOS Health Score is calculated across five equally-weighted dimensions:
|
| 138 |
+
|
| 139 |
+
```
|
| 140 |
+
Health Score = (Revenue Strength × 20%) +
|
| 141 |
+
(Customer Retention × 20%) +
|
| 142 |
+
(Market Position × 20%) +
|
| 143 |
+
(Technology Adoption × 20%) +
|
| 144 |
+
(Growth Trajectory × 20%)
|
| 145 |
+
|
| 146 |
+
Where each dimension is scored 0–100.
|
| 147 |
+
Maximum Score: 100
|
| 148 |
+
```
|
| 149 |
+
|
| 150 |
+
| Score Range | Label | Interpretation |
|
| 151 |
+
|------------|-------|---------------|
|
| 152 |
+
| 80 – 100 | 🟢 Excellent | Best-in-class. Scale aggressively. |
|
| 153 |
+
| 65 – 79 | 🔵 Good | Strong foundation. Focus on 1–2 gaps. |
|
| 154 |
+
| 45 – 64 | 🟡 Fair | Visible weaknesses. Targeted fixes needed. |
|
| 155 |
+
| 30 – 44 | 🟠 Below Average | Systemic issues. Restructure required. |
|
| 156 |
+
| 0 – 29 | 🔴 Critical | Immediate intervention. Prioritise survival. |
|
| 157 |
+
|
| 158 |
+
---
|
| 159 |
+
|
| 160 |
+
### Intended Use
|
| 161 |
+
|
| 162 |
+
#### ✅ Primary Use Cases
|
| 163 |
+
|
| 164 |
+
- **Myanmar Gold Shops & Jewellers** — the lifeblood of Myanmar's retail economy, underserved by digital tools
|
| 165 |
+
- **Fashion & Apparel SMEs** — fast-moving businesses in Yangon, Mandalay, Naypyidaw
|
| 166 |
+
- **F&B Operators** — restaurants, tea shops, catering businesses
|
| 167 |
+
- **Cosmetics & Beauty Brands** — direct-to-consumer Myanmar brands scaling up
|
| 168 |
+
- **Electronics Retailers** — high-value, low-margin businesses needing operational precision
|
| 169 |
+
- **Any Myanmar SME founder** who wants strategic clarity without a consultant's fee
|
| 170 |
+
|
| 171 |
+
#### ❌ Out-of-Scope Uses
|
| 172 |
+
|
| 173 |
+
- Large corporations (BIOS is tuned for SME scale and context)
|
| 174 |
+
- Non-business tasks (general Q&A, creative writing)
|
| 175 |
+
- Legal or financial advice (BIOS provides business intelligence, not regulated advisory)
|
| 176 |
+
|
| 177 |
+
---
|
| 178 |
+
|
| 179 |
+
### How to Use
|
| 180 |
+
|
| 181 |
+
#### With the `transformers` Library
|
| 182 |
+
|
| 183 |
+
```python
|
| 184 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 185 |
+
import torch
|
| 186 |
+
|
| 187 |
+
model_id = "BIOS-kernel/BIOS-Insight-v1"
|
| 188 |
+
|
| 189 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 190 |
+
model = AutoModelForCausalLM.from_pretrained(
|
| 191 |
+
model_id,
|
| 192 |
+
torch_dtype=torch.bfloat16,
|
| 193 |
+
device_map="auto",
|
| 194 |
+
)
|
| 195 |
+
|
| 196 |
+
system_prompt = """You are BIOS — the Business Idea Operating System.
|
| 197 |
+
You are the elite AI advisor for Myanmar SMEs.
|
| 198 |
+
Always respond in valid JSON with health_score, top_3_weaknesses,
|
| 199 |
+
growth_opportunities, and priority_action_items."""
|
| 200 |
+
|
| 201 |
+
user_prompt = """Diagnose this business:
|
| 202 |
+
Business: Shwe Zin Gold & Jewellery | Industry: Gold Shop | Location: Yangon
|
| 203 |
+
Monthly Revenue: 4,200,000 MMK | Retention Rate: 28% | Team: 3 people
|
| 204 |
+
USP: Certified 99.9% pure gold with 10-year buyback guarantee
|
| 205 |
+
Pain Point: No customer follow-up system. Customers don't return.
|
| 206 |
+
12-Month Goal: 12,000,000 MMK
|
| 207 |
+
Marketing Budget: 80,000 MMK/month"""
|
| 208 |
+
|
| 209 |
+
messages = [
|
| 210 |
+
{"role": "system", "content": system_prompt},
|
| 211 |
+
{"role": "user", "content": user_prompt},
|
| 212 |
+
]
|
| 213 |
+
|
| 214 |
+
input_ids = tokenizer.apply_chat_template(
|
| 215 |
+
messages,
|
| 216 |
+
add_generation_prompt=True,
|
| 217 |
+
return_tensors="pt",
|
| 218 |
+
).to(model.device)
|
| 219 |
+
|
| 220 |
+
output = model.generate(
|
| 221 |
+
input_ids,
|
| 222 |
+
max_new_tokens=1024,
|
| 223 |
+
temperature=0.3,
|
| 224 |
+
do_sample=True,
|
| 225 |
+
pad_token_id=tokenizer.eos_token_id,
|
| 226 |
+
)
|
| 227 |
+
|
| 228 |
+
response = tokenizer.decode(
|
| 229 |
+
output[0][input_ids.shape[-1]:],
|
| 230 |
+
skip_special_tokens=True,
|
| 231 |
+
)
|
| 232 |
+
print(response)
|
| 233 |
+
```
|
| 234 |
+
|
| 235 |
+
#### With the BIOS Controller (Recommended)
|
| 236 |
+
|
| 237 |
+
```python
|
| 238 |
+
from bios_controller import BIOSController, BusinessInputs, ModelBackend
|
| 239 |
+
|
| 240 |
+
# Initialise
|
| 241 |
+
controller = BIOSController(
|
| 242 |
+
backend = ModelBackend.GROQ, # or HF_INFERENCE when BIOS-Insight-v1 is live
|
| 243 |
+
save_to_db = True, # persist to NeonDB
|
| 244 |
+
)
|
| 245 |
+
|
| 246 |
+
# Fill in the 24 business questions
|
| 247 |
+
inputs = BusinessInputs(
|
| 248 |
+
business_name = "Shwe Zin Gold & Jewellery",
|
| 249 |
+
industry = "Gold Shop",
|
| 250 |
+
location = "Yangon",
|
| 251 |
+
years_in_business = 7,
|
| 252 |
+
monthly_revenue = 4_200_000,
|
| 253 |
+
team_size = 3,
|
| 254 |
+
target_customer = "Middle-income families, 30–55, buying gold for investment and festivals",
|
| 255 |
+
acquisition_channels = ["Word-of-mouth", "Facebook", "Walk-in"],
|
| 256 |
+
avg_customer_lifetime_value= 350_000,
|
| 257 |
+
retention_rate = 28.0,
|
| 258 |
+
main_competitors = "Dagon Gold, KBZ Gems",
|
| 259 |
+
unique_selling_proposition = "Certified 99.9% gold. Transparent pricing. 10-year buyback guarantee.",
|
| 260 |
+
sales_channels = ["Physical Store", "Facebook"],
|
| 261 |
+
operational_challenge = "Inventory management",
|
| 262 |
+
biggest_pain_point = "No system to follow up with customers after first purchase.",
|
| 263 |
+
current_technology = ["Spreadsheets"],
|
| 264 |
+
marketing_channels = ["Facebook", "Word-of-mouth"],
|
| 265 |
+
monthly_marketing_budget = 80_000,
|
| 266 |
+
goal_3_month = 5_500_000,
|
| 267 |
+
goal_6_month = 7_000_000,
|
| 268 |
+
goal_12_month = 12_000_000,
|
| 269 |
+
budget_constraint = "Tight (50-200K)",
|
| 270 |
+
tech_readiness = "Somewhat ready",
|
| 271 |
+
preferred_language = "English",
|
| 272 |
+
)
|
| 273 |
+
|
| 274 |
+
# Run the full diagnosis pipeline
|
| 275 |
+
report = controller.run_diagnosis(inputs)
|
| 276 |
+
|
| 277 |
+
# Access structured results
|
| 278 |
+
print(f"Health Score : {report.health_score}/100 ({report.health_label})")
|
| 279 |
+
print(f"Top Weakness : {report.top_3_weaknesses[0].label}")
|
| 280 |
+
print(f"Best Opportunity : {report.growth_opportunities[0].title}")
|
| 281 |
+
print(f"\nAI Narrative:\n{report.ai_narrative}")
|
| 282 |
+
|
| 283 |
+
# Full JSON output
|
| 284 |
+
print(report.to_json())
|
| 285 |
+
```
|
| 286 |
+
|
| 287 |
+
#### With HuggingFace Inference API
|
| 288 |
+
|
| 289 |
+
```python
|
| 290 |
+
from huggingface_hub import InferenceClient
|
| 291 |
+
|
| 292 |
+
client = InferenceClient(
|
| 293 |
+
model = "BIOS-kernel/BIOS-Insight-v1",
|
| 294 |
+
token = "hf_your_token_here",
|
| 295 |
+
)
|
| 296 |
+
|
| 297 |
+
response = client.chat_completion(
|
| 298 |
+
messages=[
|
| 299 |
+
{"role": "system", "content": "You are BIOS. Respond in JSON."},
|
| 300 |
+
{"role": "user", "content": "Diagnose: Gold Shop, 4.2M MMK revenue, 28% retention."},
|
| 301 |
+
],
|
| 302 |
+
max_tokens = 1024,
|
| 303 |
+
temperature = 0.3,
|
| 304 |
+
)
|
| 305 |
+
print(response.choices[0].message.content)
|
| 306 |
+
```
|
| 307 |
+
|
| 308 |
+
---
|
| 309 |
+
|
| 310 |
+
### Switching Models (Base vs Fine-tuned)
|
| 311 |
+
|
| 312 |
+
```python
|
| 313 |
+
controller = BIOSController(backend=ModelBackend.GROQ)
|
| 314 |
+
|
| 315 |
+
# Use base LLaMA-3.3-70B (default, available now)
|
| 316 |
+
report_base = controller.run_diagnosis(inputs)
|
| 317 |
+
|
| 318 |
+
# Switch to BIOS-Insight-v1 once published on HuggingFace
|
| 319 |
+
controller.switch_to_bios_insight()
|
| 320 |
+
report_bios = controller.run_diagnosis(inputs)
|
| 321 |
+
|
| 322 |
+
# Switch back to base
|
| 323 |
+
controller.switch_to_base()
|
| 324 |
+
```
|
| 325 |
+
|
| 326 |
+
---
|
| 327 |
+
|
| 328 |
+
### NeonDB Integration
|
| 329 |
+
|
| 330 |
+
```python
|
| 331 |
+
import os
|
| 332 |
+
os.environ["DATABASE_URL"] = "postgresql://user:pass@ep-xxx.neon.tech/neondb?sslmode=require"
|
| 333 |
+
|
| 334 |
+
controller = BIOSController(save_to_db=True)
|
| 335 |
+
report = controller.run_diagnosis(inputs)
|
| 336 |
+
|
| 337 |
+
# Retrieve saved report
|
| 338 |
+
saved = controller.get_report(report.session_id)
|
| 339 |
+
|
| 340 |
+
# List all diagnoses
|
| 341 |
+
history = controller.list_reports(limit=10)
|
| 342 |
+
```
|
| 343 |
+
|
| 344 |
+
---
|
| 345 |
+
|
| 346 |
+
### Limitations
|
| 347 |
+
|
| 348 |
+
- Benchmarks are calibrated for Myanmar market (MMK currency, Yangon/Mandalay/Naypyidaw context)
|
| 349 |
+
- Growth projections are estimates, not guarantees — market conditions vary
|
| 350 |
+
- The model does not access real-time data or the internet
|
| 351 |
+
- Legal and financial decisions should always be reviewed by qualified professionals
|
| 352 |
+
|
| 353 |
+
---
|
| 354 |
+
|
| 355 |
+
### Citation
|
| 356 |
+
|
| 357 |
+
```bibtex
|
| 358 |
+
@misc{bios-insight-v1,
|
| 359 |
+
title = {BIOS-Insight-v1: Business Idea Operating System for Myanmar SMEs},
|
| 360 |
+
author = {BIOS-kernel},
|
| 361 |
+
year = {2026},
|
| 362 |
+
howpublished = {\url{https://huggingface.co/BIOS-kernel/BIOS-Insight-v1}},
|
| 363 |
+
note = {Fine-tuned on LLaMA 3.3 70B Instruct for Myanmar business diagnostics}
|
| 364 |
+
}
|
| 365 |
+
```
|
| 366 |
+
|
| 367 |
+
---
|
| 368 |
+
|
| 369 |
+
---
|
| 370 |
+
|
| 371 |
+
## 🇲🇲 မြန်မာဘာသာ (Burmese)
|
| 372 |
+
|
| 373 |
+
### မော်ဒယ်ဖော်ပြချက်
|
| 374 |
+
|
| 375 |
+
**BIOS-Insight-v1** သည် **LLaMA 3.3 70B Instruct** ကို အခြေခံ၍ fine-tune ပြုလုပ်ထားသော AI မော်ဒယ်တစ်ခုဖြစ်ပြီး၊ မြန်မာနိုင်ငံ၏ SME (အသေးစားနှင့် အလတ်စားလုပ်ငန်းများ) — ရွှေဆိုင်များ၊ ဖက်ရှင်ဆိုင်များ၊ စားသောက်ဆိုင်များ၊ နှင့် နောင်လာမည့် Southeast Asia ၏ လုပ်ငန်းရှင်များအတွက် **Business Idea Operating System (BIOS)** ၏ AI အဓိကအင်ဂျင်အဖြစ် ဒီဇိုင်းထုတ်ထားသည်။
|
| 376 |
+
|
| 377 |
+
BIOS သည် chatbot တစ်ခုမဟုတ်ပါ။ ၎င်းသည် **သင်၏လုပ်ငန်းအကြံဥာဏ်များအတွက် Operating System** တစ်ခုဖြစ်သည် — Windows က သင်၏ကွန်ပျူတာကို run သကဲ့သို့၊ BIOS က သင်၏လုပ်ငန်းဗျူဟာကို run သည်။ မေးထားသောမေးခွန်းတိုင်း၊ ဖော်ထုတ်သော အားနည်းချက်တိုင်း၊ အဆင့်သတ်မှတ်ထားသော အခွင့်အလမ်းတိုင်း — ဆောင်ရွက်မှုအားလုံးကို AI kernel တစ်ခုတည်းဖြင့် လမ်းညွှန်သည်။
|
| 378 |
+
|
| 379 |
+
---
|
| 380 |
+
|
| 381 |
+
### ရည်ရွယ်သောအသုံးပြုနယ်ပယ်
|
| 382 |
+
|
| 383 |
+
BIOS-Insight-v1 ကို အောက်ပါလုပ်ငန်းများအတွက် အထူးသင့်တော်သည်:
|
| 384 |
+
|
| 385 |
+
**✅ အဓိကအသုံးပြုနယ်ပယ်များ**
|
| 386 |
+
|
| 387 |
+
- 🥇 **မြန်မာရွှေဆိုင်များနှင့် လက်ဝတ်ရတနာဆိုင်များ** — မြန်မာ့လက်လီကုန်ခြောက်စီးပွားရေး၏ အသက်ကြောဖြစ်သော ဆိုင်များ
|
| 388 |
+
- 👗 **ဖက်ရှင်နှင့် အဝတ်အထည် SME များ** — ရန်ကုန်၊ မန္တလေး၊ နေပြည်တော်ရှိ ဆိုင်များ
|
| 389 |
+
- 🍜 **F&B လုပ်ငန်းများ** — စားသောက်ဆိုင်၊ လက်ဖက်ရည်ဆိုင်၊ Catering လုပ်ငန်းများ
|
| 390 |
+
- 💄 **လှပရေးနှင့် ကောင်မီတစ်ဆ Brand များ** — မြန်မာ DTC Brand များ
|
| 391 |
+
- 📱 **Electronics ဆိုင်များ** — ကုန်ပစ္စည်းတန်ဖိုးမြင့်သော၊ margin နည်းသောလုပ်ငန်းများ
|
| 392 |
+
- 🏢 **မြန်မာ SME တည်ထောင်သူများ** — consultant ဦးစောင်ကြေးမပေးဘဲ ဗျူဟာကို ရှင်းလင်းစေလိုသူများ
|
| 393 |
+
|
| 394 |
+
---
|
| 395 |
+
|
| 396 |
+
### BIOS ၏ ကျန်းမာရေးရမှတ်ဖော်မြူလာ
|
| 397 |
+
|
| 398 |
+
BIOS Health Score ကို ညီမျှသောအချိန်ချိန်ထားသော ကဏ္ဍ ၅ ခုဖြင့် တွက်ချက်သည်:
|
| 399 |
+
|
| 400 |
+
```
|
| 401 |
+
Health Score = (ဝင်ငွေခိုင်ခံ့မှု × ၂၀%) +
|
| 402 |
+
(ဖောက်သည်ဆက်လက်ဆောင်ရွက်မှု × ၂၀%) +
|
| 403 |
+
(ဈေးကွက်တွင်နေရာ × ၂၀%) +
|
| 404 |
+
(နည်းပညာဆိုင်ရာသုံးစွဲမှု × ၂၀%) +
|
| 405 |
+
(တိုးတက်မှ���ပန်းတိုင် × ၂၀%)
|
| 406 |
+
|
| 407 |
+
အမြင့်ဆုံးရမှတ်: ၁၀၀
|
| 408 |
+
```
|
| 409 |
+
|
| 410 |
+
| ရမှတ် | အမှတ်တံဆိပ် | အဓိပ္ပါယ် |
|
| 411 |
+
|------|------------|---------|
|
| 412 |
+
| ၈၀–၁၀၀ | 🟢 ထူးခြားကောင်းမွန်သော | ကဏ္ဍ အကောင်းဆုံး။ တိုးချဲ့ပါ။ |
|
| 413 |
+
| ၆၅–၇၉ | 🔵 ကောင်းမွန်သော | ခိုင်မာသောအခြေခံ။ ကွာဟချက် ၁–၂ ခုကို အာရုံစိုက်ပါ။ |
|
| 414 |
+
| ၄၅–၆၄ | 🟡 ဖြစ်နိုင်သော | မြင်သာသောအားနည်းချက်များ။ ပစ်မှတ်ထားပြင်ဆင်ရန်လိုသည်။ |
|
| 415 |
+
| ၃၀–၄၄ | 🟠 ပျမ်းမျှအောက် | စနစ်ဆိုင်ရာပြဿနာများ။ ပြန်ဖွဲ့စည်းရန်လိုသည်။ |
|
| 416 |
+
| ၀–၂၉ | 🔴 အရေးပေါ် | ချက်ချင်းဝင်ရောက်ကူညီရန်လိုသည်။ |
|
| 417 |
+
|
| 418 |
+
---
|
| 419 |
+
|
| 420 |
+
### မည်သို့အသုံးပြုမည်နည်း (`transformers` နှင့်)
|
| 421 |
+
|
| 422 |
+
```python
|
| 423 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 424 |
+
import torch
|
| 425 |
+
|
| 426 |
+
model_id = "BIOS-kernel/BIOS-Insight-v1"
|
| 427 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 428 |
+
model = AutoModelForCausalLM.from_pretrained(
|
| 429 |
+
model_id,
|
| 430 |
+
torch_dtype = torch.bfloat16,
|
| 431 |
+
device_map = "auto",
|
| 432 |
+
)
|
| 433 |
+
|
| 434 |
+
# မြန်မာဘာသာဖြင့် မေးမြန်းနိုင်သည်
|
| 435 |
+
messages = [
|
| 436 |
+
{
|
| 437 |
+
"role": "system",
|
| 438 |
+
"content": (
|
| 439 |
+
"သင်သည် BIOS ဖြစ်သည် — Business Idea Operating System။ "
|
| 440 |
+
"မြန်မာ SME များအတွက် elite AI အကြံပေး။ "
|
| 441 |
+
"JSON ဖော်မတ်ဖြင့် ဖြေပါ။"
|
| 442 |
+
),
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"role": "user",
|
| 446 |
+
"content": (
|
| 447 |
+
"ဤလုပ်ငန်းကို စစ်ဆေးပါ:\n"
|
| 448 |
+
"လုပ်ငန်း: ရွှေဇင် ရွှေနှင့် လက်ဝတ်ရတနာ | ကဏ္ဍ: ရွှေဆိုင် | တည်နေရာ: ရန်ကုန်\n"
|
| 449 |
+
"လစဉ်ဝင်ငွေ: ၄,၂၀၀,၀၀၀ ကျပ် | Retention Rate: ၂၈% | အဖွဲ့ဝင်: ၃ ဦး\n"
|
| 450 |
+
"အကြီးဆုံးပြဿနာ: ဖောက်သည်များကို ပြန်မလာအောင် ဆက်သွယ်နိုင်သောစနစ် မရှိ\n"
|
| 451 |
+
"၁၂ လပန်းတိုင်: ၁၂,၀၀၀,၀၀၀ ကျပ်"
|
| 452 |
+
),
|
| 453 |
+
},
|
| 454 |
+
]
|
| 455 |
+
|
| 456 |
+
input_ids = tokenizer.apply_chat_template(
|
| 457 |
+
messages, add_generation_prompt=True, return_tensors="pt"
|
| 458 |
+
).to(model.device)
|
| 459 |
+
|
| 460 |
+
output = model.generate(
|
| 461 |
+
input_ids, max_new_tokens=1024, temperature=0.3, do_sample=True,
|
| 462 |
+
pad_token_id=tokenizer.eos_token_id,
|
| 463 |
+
)
|
| 464 |
+
response = tokenizer.decode(output[0][input_ids.shape[-1]:], skip_special_tokens=True)
|
| 465 |
+
print(response)
|
| 466 |
+
```
|
| 467 |
+
|
| 468 |
+
---
|
| 469 |
+
|
| 470 |
+
### BIOS Controller ဖြင့် အသုံးပြုခြင်း
|
| 471 |
+
|
| 472 |
+
```python
|
| 473 |
+
from bios_controller import BIOSController, BusinessInputs, ModelBackend
|
| 474 |
+
|
| 475 |
+
controller = BIOSController(backend=ModelBackend.GROQ, save_to_db=True)
|
| 476 |
+
|
| 477 |
+
inputs = BusinessInputs(
|
| 478 |
+
business_name = "ရွှေဇင် ရွှေနှင့် လက်ဝတ်ရတနာ",
|
| 479 |
+
industry = "Gold Shop",
|
| 480 |
+
location = "ရန်ကုန်",
|
| 481 |
+
years_in_business = 7,
|
| 482 |
+
monthly_revenue = 4_200_000,
|
| 483 |
+
team_size = 3,
|
| 484 |
+
retention_rate = 28.0,
|
| 485 |
+
unique_selling_proposition = "အသိအမှတ်ပြုထားသော ၉၉.၉% ရွှေစစ် — ၁၀ နှစ် buyback အာမခံ",
|
| 486 |
+
biggest_pain_point = "ဖောက်သည်များကို ပထမဝယ်ပြီးနောက် ဆက်သွယ်နိုင်သောစနစ် မရှိ",
|
| 487 |
+
goal_12_month = 12_000_000,
|
| 488 |
+
preferred_language = "မြန်မာဘာသာ",
|
| 489 |
+
# ... (မေးခွန်း ၂၄ ခုလုံး)
|
| 490 |
+
)
|
| 491 |
+
|
| 492 |
+
report = controller.run_diagnosis(inputs)
|
| 493 |
+
print(f"ကျန်းမာရေးရမှတ်: {report.health_score}/၁၀၀ ({report.health_label})")
|
| 494 |
+
print(f"AI အစီရင်ခံချက်:\n{report.ai_narrative}")
|
| 495 |
+
```
|
| 496 |
+
|
| 497 |
+
---
|
| 498 |
+
|
| 499 |
+
### လုံခြုံရေးနှင့် ကန့်သတ်ချက်များ
|
| 500 |
+
|
| 501 |
+
- Benchmark များသည် မြန်မာ့ဈေးကွက်အခြေအနေ (MMK ငွေကြေး) အတွက် ချိန်ညှိထားသည်
|
| 502 |
+
- ကြီးထွားမှုခန့်မှန်းချက်များသည် estimate များသာဖြစ်ပြီး အာမခံချက်မပေးနိုင်ပါ
|
| 503 |
+
- ဥပဒေနှင့် ဘဏ္ဍာရေးဆိုင်ရာ ဆုံးဖြတ်ချက်များကို အရည်အချင်းပြည့်ဝသောကျွမ်းကျင်သူများနှင့် ပြန်လည်စစ်ဆေးသင့်သည်
|
| 504 |
+
|
| 505 |
+
---
|
| 506 |
+
|
| 507 |
+
<div align="center">
|
| 508 |
+
|
| 509 |
+
**BIOS — Business Idea Operating System**
|
| 510 |
+
|
| 511 |
+
*"သင်၏လုပ်ငန်းကို ကျွန်ုပ်တို့ ရိုးရိုးစစ်ဆေးတာမဟုတ်ပါ။ ကျွန်ုပ်တို့ ၎င်းကို လင်းထိန်စေသည်။"*
|
| 512 |
+
|
| 513 |
+
*"We don't just analyse businesses. We illuminate them."*
|
| 514 |
+
|
| 515 |
+
[](https://huggingface.co/BIOS-kernel/BIOS-Insight-v1)
|
| 516 |
+
|
| 517 |
+
</div>
|