Wordtoken exposes a focused API for Myanmar word segmentation, POS tagging, and batch inference. The service is live at wordtoken.ygn.app and fronts a Hugging Face-backed XLM-RoBERTa + BiLSTM + CRF model behind Caddy HTTPS.
Use this landing page for fast onboarding, and open the wiki for operational notes, deployment topology, and release guidance.
Returns a list of token strings for a single input sentence.
curl -X POST https://wordtoken.ygn.app/api/v1/segment \
-H 'content-type: application/json' \
-d '{"text":"ကျွန်တော်သည်ကျောင်းသွားသည်"}'
Returns token/POS pairs in one call for downstream NLP apps.
curl -X POST https://wordtoken.ygn.app/api/v1/pos \
-H 'content-type: application/json' \
-d '{"text":"ကျွန်တော်သည်ကျောင်းသွားသည်"}'
Processes multiple strings per request with batch-size controls.
curl -X POST https://wordtoken.ygn.app/api/v1/batch \
-H 'content-type: application/json' \
-d '{"texts":["မင်္ဂလာပါ","hello world"]}'
The API is HTTPS-first. Plain HTTP requests are redirected by
Caddy, and the OpenAPI UIs stay available at /docs
and /redoc.
curl https://wordtoken.ygn.app/health
curl -X POST https://wordtoken.ygn.app/api/v1/segment \
-H 'content-type: application/json' \
-d '{"text":"မြန်မာဘာသာသည်လှပသောဘာသာတစ်ခုဖြစ်သည်"}'
{
"input": "ကျွန်တော်သည်ကျောင်းသွားသည်",
"words": [
"ကျွန်တော်",
"သည်",
"ကျောင်း",
"သွား",
"သည်"
],
"processing_time_ms": 105.12
}
| Resource | Purpose |
|---|---|
| /wiki | Operations handbook, deployment topology, and CI/CD notes. |
| /docs | Swagger UI backed by the live OpenAPI schema. |
| /redoc | Alternative API reference with a long-form layout. |
| GitHub repository | Source code, deploy workflow, and infrastructure templates. |