Een FastAPI-implementatie Exact kopiëren van de API van EasyNMT (https://github.com/UKPLAB/EasyNMT) een uitstekend maar verlaten neuraal-machine-vertalingsproject.
Maar ik heb zo veel leuke functies toegevoegd om de betrouwbaarheid te verhogen en het klaar te maken voor gebruik in een productiesysteem.Denk snelle vertaling zelf gehost...).
Sinds het begin van deze blog, een grote passie is auto-vertaling van blog artikelen.
JA Ik weet dat 'google doet dit' in browsers etc..etc...maar dat is niet het punt.mostlylucid-nmtIk wilde weten hoe ik het moest doen!
Plus het is leuk om gastvrij te zijn voor mensen die geen Engels lezen (ook al lezen ze Engels als tweede taal, het is FAR moeilijker om te ontleden).Dus bedacht ik hoe het te doen; evenals het delen van hoe dit soort systeem te bouwen.Oh en het gaf me ideeën over hoe het te gebruiken in ASP.NET voor automatische lokalisatie van tekst (inclusief dynamische tekst) met behulp van SignalR & een glad realtime updatesysteem. (
Blijf kijken.Oh and I've made a demo available here; https://nmtdemo.mostlylucid.net/demo/ it's only running in an old laptop without no GPU but gives you the idea (and lets me test lengevity).In wezen; mensen schrijven onzin tekst die SUPER luidruchtig is voor machines om efficiënt om te gaan.
Er werd dus veel uitgezocht over problemen met EasyNMT (het was echt een onderzoeksproject).
Ischreef een heel systeemhttp://<server>:<port>/demoom dat te laten gebeuren met een geweldig project genaamd EasyNMT.
Het is een eenvoudige, snelle manier om een vertaling te krijgen API zonder de noodzaak om te betalen voor een bepaalde dienst of een full-size LLM om vertaling te krijgen (langzaam).
**Maar naarmate de tijd voortging, begonnen de scheuren te zien.**Het was tijd voor iets beters.
**Zoals gewoonlijk is het allemaal op GitHub en alle gratis voor gebruik etc...**Docker Pulls
Reusing loaded model for en->de (3/10 models in cache)Need to load model for en->fr (3/10 models in cache)Belangrijke updates (v3.1) - Intelligentie & ZichtbaarheidNieuw in v3.1:
====================================================================================================
🚀 DOWNLOADING MODEL
Model: facebook/mbart-large-50-many-to-many-mmt
Family: mbart50
Direction: en → bn
Device: GPU (cuda:0)
Total Size: 2.46 GB
Files: 6 main files
====================================================================================================
[Progress bars for each file...]
====================================================================================================
✅ MODEL READY
Model: facebook/mbart-large-50-many-to-many-mmt
Translation: en → bn is now available
====================================================================================================
**: Standaard cachegrootte gestoten naar 10 modellen (vanaf 6)**Per model apparaatlogging
[Pivot] Languages reachable from en: 85 languages
[Pivot] Languages that can reach bn: 42 languages
[Pivot] Found 38 possible pivot languages
[Pivot] Selected pivot: en → hi → bn (both legs verified)
: Toont doelapparaat (GPU/CPU) in bannerVoortgangsbalk
Request: en→bn with opus-mt
Trying families: ['opus-mt', 'mbart50', 'm2m100'] ✓ All three!
opus-mt: Failed (model doesn't exist)
mbart50: Success! (auto-fallback worked)
Data-aangedreven intelligente pivot-selectie- Geen blinde pogingen meer.
Loading mbart50 model on GPU (cuda:0)Model loaded on device: cuda:0Successfully loaded... on GPU (cuda:0): Zal niet proberen en→es→bn als es→bn niet bestaatVoorbeeld voor nl→bn
en->hi, hi->bn[Pivot] Both legs loaded and cached. Ready to translate.: Zie precies waarom elk draaipunt werd gekozen of overgeslagen4.
model_familyProbeert altijd terugvallen: Geen twee keer hetzelfde model meer proberenVoorbeeldstroom
**Succesboodschap omvat:**6.
**- Werkt al in demo:**Demo dropdown maakt het selecteren van opus-mt, mbart50 of m2m100 mogelijk
requirements-prod.txtVerbeterde demopagina**- Productie-ready interactieve interface:**Volledige viewport layout (100vw/100vh) voor meeslepende vertaalervaring
**: FP16 ingeschakeld, BATCH_SIZE=64, MAX_INFLIGHT=1 (optimale voor enkele GPU)**CPU
**- Kleinere, snellere afbeeldingen:**Verwijderde test afhankelijkheden (pytest, pytest-cov) uit productiebouw
Uitgebreide test & belastingstest- Valideer alles:
met realistische verkeerspatronenCross-platform validatiescripts
/discover/opus-mt(PowerShell + Bash)/discover/mbart50Tests voor modeldownloads en spilvertaling terugval/discover/m2m100Geautomatiseerde rooktests voor snelle validatie**5.**Implementatiedocumentatie
Kubernetes manifesteert zich met PVC, grondstoffenlimieten, gezondheidscontrolesVoorbeelden van Azure Container Instances
scottgal/mostlylucid-nmt:cpuLaadbegeleidings- en monitoringaanbevelingen:latestConcurrency versus throughput trade-offs uitgelegdscottgal/mostlylucid-nmt:cpu-min6.scottgal/mostlylucid-nmt:gpuDrie modelfamiliesscottgal/mostlylucid-nmt:gpu-min- Kies het beste voor uw behoeften:Opus-MT: 1200+ paar, beste kwaliteit (afzonderlijke modellen)
latest, min, gpu, gpu-min: 50 talen, enkel 2.4GB model, 2.450 paren20250108.143022: 100 talen, enkele 2.2GB model, 9.900 paarAutomatisch terugvallen- Intelligent selecteert het best beschikbare model:
- Alle mBART50 paren- Alle M2M100 paren
Geen vooraf geladen modellen (download on-demand)
Schakel modelfamilies om zonder herbouw**10.**Single Docker repository
cpu(oflatest) | scottgal/mostlylucid-nmt:cpu) - CPU
| cpu-min | scottgal/mostlylucid-nmt:cpu-min- CPU minimaal
| gpu | scottgal/mostlylucid-nmt:gpu- GPU met CUDA 12.6
| gpu-min | scottgal/mostlylucid-nmt:gpu-min- GPU minimaal**11.**Juiste versie
Volledige OCI labels voor het volgen van versies, bouwdata en git commits
docker run -d \
--name mostlylucid-nmt \
-p 8000:8000 \
scottgal/mostlylucid-nmt
12.
curl -X POST "http://localhost:8000/translate" \
-H "Content-Type: application/json" \
-d '{
"text": ["Hello, how are you?"],
"target_lang": "de"
}'
Nieuwste basisafbeeldingen
{
"translated": ["Hallo, wie geht es Ihnen?"],
"target_lang": "de",
"source_lang": "en",
"translation_time": 0.34
}
Python 3.12-slank
docker run -d \
--name mostlylucid-nmt \
--gpus all \
-p 8000:8000 \
-e EASYNMT_MODEL_ARGS='{"torch_dtype":"fp16"}' \
scottgal/mostlylucid-nmt:gpu
CUDA 12.6
met Ubuntu 24.04 voor GPU beelden (laatste NVIDIA stack)
docker run -d \
--name mostlylucid-nmt \
-p 8000:8000 \
-v $HOME/model-cache:/models \
-e MODEL_CACHE_DIR=/models \
scottgal/mostlylucid-nmt:cpu-min
PyTorch met CUDA 12.4
docker run -d `
--name mostlylucid-nmt `
-p 8000:8000 `
-v ${HOME}/model-cache:/models `
-e MODEL_CACHE_DIR=/models `
scottgal/mostlylucid-nmt:cpu-min
(compatibel met CUDA 12,6 runtime)
docker run -d ^
--name mostlylucid-nmt ^
-p 8000:8000 ^
-v %USERPROFILE%/model-cache:/models ^
-e MODEL_CACHE_DIR=/models ^
scottgal/mostlylucid-nmt:cpu-min
Alle afhankelijkheden bijgewerkt naar de nieuwste beveiligde versies
13.
curl http://localhost:8000/healthz
Deprecatiewaarschuwingen (fixed deprecation warnings)
Verwijderde verouderde TRANSFORMERS_CACHE (nu met HF_HOME)Compatibel met Transformers v5Snel starten (5 minuten)
http://localhost:8000/demo/
Hier is de absolute eenvoudigste manier om te draaien meestal lucid-nmt:
Beschikbare Docker-afbeeldingen
Tag: volledige afbeeldingsnaam Grootte Beschrijving Gebruikscase:
Minimale afbeeldingen
De grootte van de container klein houden
Vereist NVIDIA Docker runtime:
Gezondheidscontrole:
// Example: Translating a 5000-word article
Input: Long article with multiple paragraphs
Step 1: Split by paragraphs (preserves structure)
→ Paragraph 1 (800 chars)
→ Paragraph 2 (1200 chars)
→ Paragraph 3 (600 chars)
...
Step 2: Group into ~1000 character chunks
→ Chunk 1: Paragraphs 1-2
→ Chunk 2: Paragraph 3-4
→ Chunk 3: Paragraphs 5-6
Step 3: Translate each chunk sequentially
→ Shows progress: "Translating chunk 1/3..."
→ Shows progress: "Translating chunk 2/3..."
→ Shows progress: "Translating chunk 3/3..."
Step 4: Reassemble with paragraph breaks
→ Final output: Complete translated article with preserved formatting
Auto-bevolkte taal dropdowns van de live service
Behandelt automatisch grote tekstinvoeren van elke grootte
3.
Geavanceerde opties:
Splitsing van de straf:
Real-Time Statistieken:
: Idle → Translating → Klaar/Fout
: 100 talen, 9.900 paar/demo/Zie precies welke taalparen beschikbaar zijn voordat u vertaalt
De demo implementeert intelligente tekstchunking aan de clientzijde:Waarom de Demo gebruiken?Snel testenTestvertalingen zonder code te schrijvenBeschikbaarheid taalpaar valideren
Vergelijk vertaalkwaliteit met verschillende bundelgroottes
Volledige blogpost plakken (5000+ woorden)Demo brokken het automatisch in beheersbare stukkenToont vooruitgang zoals elke brok vertaalt
TaaldetectieTekst in onbekende taal plakkenKlik op "Taal detecteren"
**Geen tegendruk of wachtrij.**Stuur te veel verzoeken en het valt gewoon over.MODEL_FAMILYGeen opmerkzaamheid.
# Opus-MT (default, best quality)
MODEL_FAMILY=opus-mt
# mBART50 (50 languages, single model)
MODEL_FAMILY=mbart50
# M2M100 (100 languages, broadest coverage)
MODEL_FAMILY=m2m100
De oplossing: Meestal Lucid-NMTDus... besloot ik om een nieuwe en verbeterde EasyNMT te bouwen, numeestal lucid-nmt
MODEL_FAMILYDit is wat het beter maakt:opus-mtOndersteuning van de familie met meerdere modelsOpus-MT (Helsinki-NLP) - Standaard
# Set primary to Opus-MT (best quality)
MODEL_FAMILY=opus-mt
AUTO_MODEL_FALLBACK=1
MODEL_FALLBACK_ORDER=opus-mt,mbart50,m2m100
# Request Ukrainian → French
# 1. Try Opus-MT first (not available)
# 2. Automatically fall back to mBART50 (available!)
# 3. Translation succeeds with mBART50
Dekking:
300-500MB per richting
# Enable auto-fallback (default: enabled)
AUTO_MODEL_FALLBACK=1
# Set fallback priority (default: opus-mt → mbart50 → m2m100)
MODEL_FALLBACK_ORDER="opus-mt,mbart50,m2m100"
# Disable for strict single-family mode
AUTO_MODEL_FALLBACK=0
Voorbeeld:
-minModelgrootte:Voordeel:
Schakelen is eenvoudig
Een van de krachtigste nieuwe features is
**Hoe het werkt:**Je hebt een primaire ingesteld.
Voordelen:
Steun 100+ talen zonder meerdere implementaties te beheren
Transparant loggen:
Meervoudige gezinsondersteuning
# NMT: Fits on a USB stick
du -sh model-cache/
2.5G model-cache/
# LLM: Needs serious storage
du -sh llama-models/
140G llama-models/
Modelbevindingseindpunten
Symbolen?
LRU-modelcaching
EasyNMT-compatibele API
varianten met een volume-kaart cache voor kleinere implementaties.
Waarom NMT over LLM's voor vertaling?
Input: "The API returns a 429 status code when rate limited."
NMT (Opus-MT): "Die API gibt einen 429-Statuscode zurück, wenn sie ratenbegrenzt ist."
(Accurate, preserves technical terms)
LLM (might do): "Die API sendet den Fehlercode 429, wenn zu viele Anfragen gestellt werden."
(Interprets rather than translates, adds context not in original)
Hier is de reality check gebaseerd op productiegebruik:
GPT-4: 3-10 seconden per aanvraag (API latentie + generatie)
GPT-4 API**: 30-60 seconden**Lokale Llama 70B
Opus-MT (per richting): 300-500MB
MBART50 (alle 50 talen): 2.4GBM2M100 (alle 100 talen): 2.2GB
flowchart LR
A[HTTP Client] --> B[API Gateway]
B --> C[Translation Endpoint]
C --> D{Has Capacity?}
D -->|Yes| E[Translation Service]
D -->|No| F[Queue with 429]
F --> E
E --> G[Process Pipeline]
G --> H[Get Model from Cache]
H --> I[Translate]
I --> J[Return Response]
J --> A
LLM's:
Retry-AfterKosten: $10-20/maand VPS**NMT sterktes:**Speciaal voor vertaling opgeleidRetry-AfterConsistente kwaliteit (zelfde input = dezelfde output)
Geen prompt engineering nodig
sequenceDiagram
participant Client
participant API
participant Queue
participant Translator
participant Cache
participant Model
Client->>API: POST /translate
API->>Queue: Acquire slot
alt Queue has space
Queue-->>API: Slot acquired
API->>Translator: Process translation
Translator->>Translator: Sanitize input
Translator->>Translator: Split sentences
Translator->>Translator: Chunk text
Translator->>Translator: Mask symbols
Translator->>Cache: Get model (en→de)
alt Cache hit
Cache-->>Translator: Return cached model
else Cache miss
Cache->>Model: Load from Hugging Face
Model-->>Cache: Pipeline loaded
Cache->>Cache: Evict old if at capacity
Cache-->>Translator: Return model
end
Translator->>Model: Translate batches
Model-->>Translator: Translations
Translator->>Translator: Unmask symbols
Translator->>Translator: Post-process
Translator-->>API: Translations
API->>Queue: Release slot
API-->>Client: 200 OK + translations
else Queue full
Queue-->>API: Overflow error
API-->>Client: 429 Too Many Requests\nRetry-After: X seconds
end
Voorbeeldscenario:
graph LR
A[Raw Input] --> B{Sanitize?}
B -->|Yes| C[Check Noise]
B -->|No| D[Split Sentences]
C -->|Is Noise| Z[Return Placeholder]
C -->|Valid| D
D --> E[Enforce Max Length]
E --> F[Chunk for Batching]
F --> G{Symbol Masking?}
G -->|Yes| H[Mask Digits/Punct/Emoji]
G -->|No| I[Translate]
H --> I
I --> J{Direct Model?}
J -->|Available| K[Direct Translation]
J -->|Not Available| L{Pivot Fallback?}
L -->|Yes| M[src→en→tgt]
L -->|No| Z
K --> N[Unmask Syis robust input handling. Here's what happens:
**Noise Detection:**
- Strips control characters (except \t, \n, \r)
- Checks minimum character count (default: 1)
- Calculates alphanumeric ratio (default: must be ≥20%)
- Rejects pure emoji, pure punctuation, or pure whitespace
**Symbol Masking:**
Why mask symbols? Translation models are trained on text, not emoji or special symbols. These can confuse them or get mangled. So we:
1. Extract all digits, punctuation, and emoji as contiguous runs
2. Replace them with sentinel tokens: `⟪MSK0⟫`, `⟪MSK1⟫`, etc.
3. Translate the masked text
4. Restore the original symbols in their positions
Example:
Input: "Hello 👋 world! Price: $99.99" Wanneer moet u ze gebruiken? (👋) (!) (:) ($99.99)
**Post-Processing:**
After translation, we remove "symbol loops" - repeated symbols that weren't in the source:
Gebruik NMT (meestal lucid-nmt) wanneer: Je hebt consistente, snelle vertaling nodig op schaal Begrotingszaken (zelfhosting of hoog volume)
### Sentence Splitting & Chunking
Long texts get split intelligently:
```mermaid
graph TD
A[Long Text] --> B[Split on . ! ? …]
B --> C{Sentence > 500 chars?}
C -->|Yes| D[Split on word boundaries]
C -->|No| E[Keep sentence]
D --> E
E --> F[Group into chunks ≤900 chars]
F --> G[Translate each chunk]
G --> H[Join with space]
Je vertaalt technische inhoud, code, gestructureerde data
Je hebt creatieve aanpassing nodig, geen letterlijke vertaling
stateDiagram-v2
[*] --> CheckCache
CheckCache --> CacheHit: Model exists
CheckCache --> CacheMiss: Model not loaded
CacheHit --> MoveToEnd: Update LRU order
MoveToEnd --> ReturnModel
CacheMiss --> CheckCapacity
CheckCapacity --> LoadModel: Space available
CheckCapacity --> EvictOldest: Cache full
EvictOldest --> MoveToCPU: Free VRAM
MoveToCPU --> ClearCUDA: torch.cuda.empty_cache()
ClearCUDA --> LoadModel
LoadModel --> AddToCache
AddToCache --> ReturnModel
ReturnModel --> [*]
Context en culturele nuance zijn belangrijker dan snelheid
(my use case), NMT is de duidelijke winnaar:
# Semaphore limits concurrent translations
MAX_INFLIGHT = 1 # On GPU, 1 at a time for efficiency
MAX_QUEUE_SIZE = 1000 # Up to 1000 waiting
# When full:
# - Returns 429 Too Many Requests
# - Includes Retry-After header
# - Estimates wait time based on average duration
Vertaalt 100+ blogberichten naar 12 talen in ~30 minuten (GPU)
avg_duration = 2.5 seconds (tracked with EMA)
waiters = 100
slots = 1
estimated_wait = (100 / 1) * 2.5 = 250 seconds
clamped = min(250, 120) = 120 seconds
Retry-After: 120
Consistente kwaliteit voor alle posten
graph LR
A[Ukrainian Text] --> B{Direct uk→fr?}
B -->|Exists| C[Translate Directly]
B -->|Missing| D[Pivot via English]
D --> E[uk→en]
E --> F[en→fr]
F --> G[French Result]
C --> G
Totale setup: One Docker container
Het snelheidsverschil alleen al maakt NMT de enige praktische keuze voor productie vertaalleidingen.
NMT is speciaal gebouwd voor vertaling, draait op bescheiden hardware, en is 10-100x sneller dan LLM's. Als u snelle, consistente, kosteneffectieve vertaling op schaal nodig hebt, wint NMT handen naar beneden.
# src/core/cache.py
from collections import OrderedDict
import torch
class LRUPipelineCache:
"""LRU cache that automatically cleans up GPU memory when evicting models."""
def __init__(self, capacity: int):
self.cache = OrderedDict() # Maintains insertion order
self.capacity = capacity
def get(self, key: str):
"""Get model from cache, moves it to end (most recently used)."""
if key not in self.cache:
return None
self.cache.move_to_end(key) # Mark as recently used
return self.cache[key]
def put(self, key: str, value):
"""Add model to cache, evicting oldest if at capacity."""
if key in self.cache:
self.cache.move_to_end(key)
else:
self.cache[key] = value
# If cache is full, evict the oldest model
if len(self.cache) > self.capacity:
oldest_key, oldest_pipeline = self.cache.popitem(last=False)
# MAGIC: Move evicted model to CPU to free GPU memory
try:
oldest_pipeline.model.to("cpu")
if torch.cuda.is_available():
torch.cuda.empty_cache() # Tell GPU to release memory
logger.info(f"Evicted {oldest_key}, freed GPU memory")
except Exception as e:
logger.warning(f"Failed to clean GPU memory: {e}")
Overzicht architectuur
OrderedDictDe aanvraagstroom is eenvoudig:→ Verzoek gaat onmiddellijk naar de vertaaldienst
# src/services/model_manager.py
def get_pipeline(self, src: str, tgt: str):
"""Try to get translation model, with automatic fallback to other providers."""
# Determine which model families support this language pair
families_to_try = []
if config.AUTO_MODEL_FALLBACK:
# Try families in priority order: opus-mt → mbart50 → m2m100
for family in config.MODEL_FALLBACK_ORDER.split(","):
if self._is_pair_supported(src, tgt, family.strip()):
families_to_try.append(family.strip())
# Try each family until one succeeds
last_error = None
for family in families_to_try:
try:
model_name, src_lang, tgt_lang, _ = self._get_model_name_and_langs(src, tgt, family)
if family != config.MODEL_FAMILY:
logger.info(f"Using fallback '{family}' for {src}->{tgt}")
# Load the model from HuggingFace
pipeline = transformers.pipeline(
"translation",
model=model_name,
device=device_manager.device_index,
src_lang=src_lang,
tgt_lang=tgt_lang
)
self.cache.put(f"{src}->{tgt}", pipeline)
return pipeline
except Exception as e:
last_error = e
logger.warning(f"Family '{family}' failed for {src}->{tgt}: {e}")
continue # Try next family
# All families failed
raise ModelLoadError(f"{src}->{tgt}", last_error)
Nee
(LRU) verstrekt vertaalmodellen:
# src/services/queue_manager.py
import asyncio
from contextlib import asynccontextmanager
class QueueManager:
"""Manages request queuing and backpressure."""
def __init__(self, max_inflight: int, max_queue: int):
self.semaphore = asyncio.Semaphore(max_inflight) # Limit concurrent translations
self.max_queue_size = max_queue
self.waiting_count = 0
self.inflight_count = 0
self.avg_duration_sec = 5.0 # Exponential moving average
@asynccontextmanager
async def acquire_slot(self):
"""Try to get a translation slot, track metrics, handle queueing."""
# Check if queue is too full
if self.waiting_count >= self.max_queue_size:
# Calculate how long client should wait before retrying
retry_after = self._estimate_retry_after()
raise QueueOverflowError(self.waiting_count, retry_after)
self.waiting_count += 1
try:
# Wait for available slot (this is the queue!)
await self.semaphore.acquire()
self.waiting_count -= 1
self.inflight_count += 1
start_time = time.time()
yield # Let the translation happen
# Update average duration for retry-after estimates
duration = time.time() - start_time
alpha = config.RETRY_AFTER_ALPHA # Smoothing factor (0.2)
self.avg_duration_sec = alpha * duration + (1 - alpha) * self.avg_duration_sec
finally:
self.inflight_count -= 1
self.semaphore.release()
def _estimate_retry_after(self) -> int:
"""Smart calculation: how many waiting / how many slots * avg time per request."""
if self.inflight_count == 0:
return config.RETRY_AFTER_MIN_SEC
# If 10 people waiting and 2 slots available, and each takes 5 seconds:
# retry_after = (10 / 2) * 5 = 25 seconds
retry_sec = (self.waiting_count / self.semaphore._value) * self.avg_duration_sec
# Clamp between min and max
return max(
config.RETRY_AFTER_MIN_SEC,
min(int(retry_sec), config.RETRY_AFTER_MAX_SEC)
)
Cache hit → Snelle reactie
max_inflight)@asynccontextmanagerkeert terug naar clientInvoerverwerkingspijpleiding
# src/utils/symbol_masking.py
import re
def mask_symbols(text: str) -> tuple[str, dict[str, str]]:
"""Replace special symbols with placeholders before translation."""
originals = {}
masked_text = text
placeholder_counter = 0
# Pattern: Match emojis, symbols, special punctuation
# \U0001F300-\U0001F9FF = emoji range
# [\u2600-\u26FF\u2700-\u27BF] = misc symbols
symbol_pattern = re.compile(
r'[\U0001F300-\U0001F9FF\u2600-\u26FF\u2700-\u27BF'
r'\u00A9\u00AE\u2122\u2139\u3030\u303D\u3297\u3299]+'
)
for match in symbol_pattern.finditer(text):
symbol = match.group()
placeholder = f"__SYMBOL_{placeholder_counter}__"
originals[placeholder] = symbol
masked_text = masked_text.replace(symbol, placeholder, 1)
placeholder_counter += 1
return masked_text, originals
def unmask_symbols(text: str, originals: dict[str, str]) -> str:
"""Restore original symbols after translation."""
for placeholder, original in originals.items():
text = text.replace(placeholder, original)
return text
De service maakt gebruik van een geavanceerde multi-stage pijplijn om rommelige real-world tekst te verwerken:
# Before translation:
text = "Hello! 👋 Check out this cool feature 🚀"
# Mask symbols:
masked, originals = mask_symbols(text)
# masked = "Hello! __SYMBOL_0__ Check out this cool feature __SYMBOL_1__"
# originals = {"__SYMBOL_0__": "👋", "__SYMBOL_1__": "🚀"}
# Translate the masked text:
translated = translate(masked, "de") # → "Hallo! __SYMBOL_0__ Schau dir diese coole Funktion an __SYMBOL_1__"
# Unmask symbols:
final = unmask_symbols(translated, originals)
# final = "Hallo! 👋 Schau dir diese coole Funktion an 🚀"
Gemaskerd: "Hallo, MSK0 wereld, MSK1 Prijs, MSK2 en MSK3"
👋Modellen verstikken niet op enorme ingangen__SYMBOL_0__We kunnen efficiënt batchenWaarom dit belangrijk is:
# src/utils/text_processing.py
def chunk_sentences(sentences: list[str], max_chars: int = 900) -> list[list[str]]:
"""Group sentences into chunks that fit within model's max input length."""
chunks = []
current_chunk = []
current_length = 0
for sentence in sentences:
sentence_len = len(sentence)
# If this sentence alone is too long, it goes in its own chunk
if sentence_len > max_chars:
if current_chunk:
chunks.append(current_chunk)
current_chunk = []
current_length = 0
chunks.append([sentence])
continue
# If adding this sentence exceeds limit, start new chunk
if current_length + sentence_len + 1 > max_chars:
chunks.append(current_chunk)
current_chunk = [sentence]
current_length = sentence_len
else:
current_chunk.append(sentence)
current_length += sentence_len + 1 # +1 for space
# Don't forget the last chunk!
if current_chunk:
chunks.append(current_chunk)
return chunks
def split_sentences(text: str, max_sentence_chars: int = 500) -> list[str]:
"""Split text into sentences, enforcing max length."""
# Split on common sentence terminators
sentences = re.split(r'([.!?…]+\s+)', text)
result = []
for sentence in sentences:
if not sentence or sentence.isspace():
continue
# If sentence is too long, split on word boundaries
if len(sentence) > max_sentence_chars:
words = sentence.split()
current = []
current_len = 0
for word in words:
if current_len + len(word) + 1 > max_sentence_chars:
result.append(' '.join(current))
current = [word]
current_len = len(word)
else:
current.append(word)
current_len += len(word) + 1
if current:
result.append(' '.join(current))
else:
result.append(sentence.strip())
return result
GPU geheugen is kostbaar
.!?…We houden de 6 meest recente modellen warmPivot via het Engels:
# src/services/model_discovery.py
import httpx
from datetime import datetime, timedelta
class ModelDiscoveryService:
"""Discovers available translation models with 1-hour cache."""
def __init__(self):
self._cache = {} # Cache results to avoid hammering HuggingFace API
self._cache_ttl = timedelta(hours=1)
self._hf_api_base = "https://huggingface.co/api/models"
async def discover_opus_mt_pairs(self, force_refresh: bool = False):
"""Query HuggingFace for all Helsinki-NLP Opus-MT models."""
cache_key = "opus-mt"
# Check cache first
if not force_refresh and cache_key in self._cache:
cached_data, cached_time = self._cache[cache_key]
if datetime.now() - cached_time < self._cache_ttl:
return cached_data # Cache hit!
# Cache miss - query HuggingFace API
async with httpx.AsyncClient() as client:
response = await client.get(
self._hf_api_base,
params={
"author": "Helsinki-NLP",
"search": "opus-mt",
"limit": 1000
},
timeout=30.0
)
models = response.json()
# Extract language pairs from model names
# Example: "Helsinki-NLP/opus-mt-en-de" → ("en", "de")
pairs = []
for model in models:
model_id = model.get("modelId", "")
if model_id.startswith("Helsinki-NLP/opus-mt-"):
# Extract the language codes after "opus-mt-"
lang_part = model_id.replace("Helsinki-NLP/opus-mt-", "")
if "-" in lang_part:
src, tgt = lang_part.split("-", 1)
pairs.append({"source": src, "target": tgt})
# Cache the results
self._cache[cache_key] = (pairs, datetime.now())
return pairs
Deze dubbele latentie maar zorgt voor dekking voor alle ondersteunde taalparen.
httpxLaten we een aantal van de meest interessante delen van de codebase verkennen!enEen van de coolste functies is de intelligente modelcache die weet hoe GPU-geheugen te verwerken:deWat gebeurt hier?Helsinki-NLP/opus-mt-en-deGPU-opruiming
# src/core/device.py
import torch
class DeviceManager:
"""Smart device selection with GPU auto-detection."""
def __init__(self):
self.use_gpu = self._should_use_gpu()
self.device_index = self._resolve_device()
self.device_str = "cpu" if self.device_index < 0 else f"cuda:{self.device_index}"
# Auto-configure parallel translation slots based on device
if self.device_index >= 0:
# GPU: Run translations serially to avoid VRAM fragmentation
self.max_inflight = 1
else:
# CPU: Can handle multiple translations in parallel
self.max_inflight = config.MAX_WORKERS_BACKEND
self._log_device_info()
def _should_use_gpu(self) -> bool:
"""Check if GPU should be used."""
if config.USE_GPU.lower() == "false":
return False
if config.USE_GPU.lower() == "true":
return torch.cuda.is_available()
# "auto" mode: use GPU if available
return torch.cuda.is_available()
def _resolve_device(self) -> int:
"""Returns device index: -1 for CPU, 0+ for CUDA."""
if not self.use_gpu:
return -1
# Check if specific CUDA device requested
if config.DEVICE and config.DEVICE.startswith("cuda:"):
device_num = int(config.DEVICE.split(":")[1])
return device_num
return 0 # Use first GPU
def _log_device_info(self):
"""Log device information at startup."""
if self.device_index >= 0:
gpu_name = torch.cuda.get_device_name(self.device_index)
vram_gb = torch.cuda.get_device_properties(self.device_index).total_memory / 1e9
logger.info(f"Using GPU: {gpu_name} ({vram_gb:.1f}GB VRAM)")
logger.info(f"Max inflight translations: {self.max_inflight} (GPU mode)")
else:
cpu_count = os.cpu_count()
logger.info(f"Using CPU ({cpu_count} cores)")
logger.info(f"Max inflight translations: {self.max_inflight} (CPU mode)")
# Global singleton instance
device_manager = DeviceManager()
: Wanneer het uitzetten van een model, we expliciet verplaatsen naar CPU-geheugen en vertellen de GPU om zijn bronnen vrij te geven
max_inflight=1Deze slimme functie probeert meerdere AI-modelproviders automatisch als de eerste niet het taalpaar heeft dat je nodig hebt:max_inflight=4Wat gebeurt hier?DEVICE=cuda:1: Succesvolle modellen worden gecached met de taal paar sleutel
# Snippet from QueueManager showing EMA calculation
def update_avg_duration(self, new_duration: float):
"""Update average duration using exponential moving average."""
# EMA formula: new_avg = α × new_value + (1 - α) × old_avg
# α = smoothing factor (0.0 to 1.0)
# - Higher α = more weight to recent values (faster adaptation)
# - Lower α = more weight to historical values (more stable)
alpha = 0.2 # 20% weight to new value, 80% to historical
self.avg_duration_sec = (
alpha * new_duration +
(1 - alpha) * self.avg_duration_sec
)
3.
# Initial average: 5.0 seconds
# New request takes: 10.0 seconds
# EMA calculation:
new_avg = 0.2 * 10.0 + 0.8 * 5.0
= 2.0 + 4.0
= 6.0 seconds
# Next request takes: 3.0 seconds
new_avg = 0.2 * 3.0 + 0.8 * 6.0
= 0.6 + 4.8
= 5.4 seconds
Wachtrij met tegendruk aanvragen (HTTP 429)
Retry-AfterExponentieel bewegend gemiddelde: gladstrijkt pieken in de duur van de aanvraag
tijdelijk
: Vertaalmodellen soms corrupt of verwijderen emojis - dit bewaart ze perfect!
# Prefer GPU if available (default)
USE_GPU=auto
# Force GPU
USE_GPU=true
# Force CPU
USE_GPU=false
# Explicit device override
DEVICE=cuda:0
DEVICE=cpu
# Model family selection (NEW in v2.0!)
MODEL_FAMILY=opus-mt # Best quality (default)
MODEL_FAMILY=mbart50 # 50 languages, single model
MODEL_FAMILY=m2m100 # 100 languages, maximum coverage
# Auto-fallback between model families (NEW in v2.0!)
AUTO_MODEL_FALLBACK=1 # Enabled by default
MODEL_FALLBACK_ORDER="opus-mt,mbart50,m2m100" # Priority order
# Volume-mapped model cache (NEW in v2.0!)
MODEL_CACHE_DIR=/models # Persistent cache directory
# Model arguments passed to transformers.pipeline
EASYNMT_MODEL_ARGS='{"torch_dtype":"fp16"}'
EASYNMT_MODEL_ARGS='{"torch_dtype":"bf16","cache_dir":"/models"}'
# Preload models at startup (reduces first-request latency)
PRELOAD_MODELS="en->de,de->en,fr->en"
# LRU cache capacity
MAX_CACHED_MODELS=6
Breek lange teksten in stukken die passen bij modellimieten met behoud van zinsgrenzen:
**Wat gebeurt hier?**Splitsing van de straf
opus-mt: Gebruikt regex om op te splitsenmbart50met behoud van de interpunctiem2m100Hebzuchtig brokeren: Verpakt zoveel mogelijk zinnen in elke brok zonder de limiet te overschrijdenWoordgrens splitsen
1: Als één zin te lang is, splitst hij zich op spaties in plaats van middenwoord te knippen0Waarom het belangrijk is**: Vertaalmodellen hebben ingangslimieten (meestal 512-1024 tokens).**Dit zorgt ervoor dat we ze nooit overschrijden terwijl we de context intact houden.
"opus-mt,mbart50,m2m100"Async Model Discovery met Caching"m2m100,mbart50,opus-mt"Wat gebeurt hier?Async HTTP-client): Maakt niet-blokkerende HTTP-verzoeken naar HuggingFace
/models: Bewaart resultaten gedurende 1 uur om snelheidsbeperking te vermijden-v ./model-cache:/modelsen
fp16vanbf16Waarom het belangrijk isfp32: HuggingFace heeft 1200+ Opus-MT modellen.# Batch size for translation (higher = faster but more VRAM)
EASYNMT_BATCH_SIZE=16 # CPU: 8-16, GPU: 32-64
# Maximum text length per item
EASYNMT_MAX_TEXT_LEN=1000
# Maximum beam size (higher = better quality but slower)
EASYNMT_MAX_BEAM_SIZE=5
# Worker thread pools
MAX_WORKERS_BACKEND=1 # Translation workers
MAX_WORKERS_FRONTEND=2 # Language detection workers
# Enable request queueing (highly recommended)
ENABLE_QUEUE=1
# Max concurrent translations
# Auto: 1 on GPU, MAX_WORKERS_BACKEND on CPU
MAX_INFLIGHT_TRANSLATIONS=1
# Max queued requests before 429
MAX_QUEUE_SIZE=1000
# Per-request timeout (0 = disabled)
TRANSLATE_TIMEOUT_SEC=180
# Retry-After estimation
RETRY_AFTER_MIN_SEC=1 # Floor
RETRY_AFTER_MAX_SEC=120 # Ceiling
RETRY_AFTER_ALPHA=0.2 # EMA smoothing factor
# Enable input filtering
INPUT_SANITIZE=1
# Minimum alphanumeric ratio (0.2 = 20%)
INPUT_MIN_ALNUM_RATIO=0.2
# Minimum character count
INPUT_MIN_CHARS=1
# Language code for undetermined/noise
UNDETERMINED_LANG_CODE=und
# Default sentence splitting behavior
PERFORM_SENTENCE_SPLITTING_DEFAULT=1
# Max chars per sentence before word-boundary split
MAX_SENTENCE_CHARS=500
# Max chars per chunk for batching
MAX_CHUNK_CHARS=900
# Sentence joiner
JOIN_SENTENCES_WITH=" "
# Enable symbol masking
SYMBOL_MASKING=1
# What to mask
MASK_DIGITS=1 # Mask 0-9
MASK_PUNCT=1 # Mask .,!? etc.
MASK_EMOJI=1 # Mask 😀🎉 etc.
# Align response array length to input
ALIGN_RESPONSES=1
# Placeholder for failed items (when aligned)
SANITIZE_PLACEHOLDER=""
# Response format
EASYNMT_RESPONSE_MODE=strings # ["translation1", "translation2"]
EASYNMT_RESPONSE_MODE=objects # [{"text":"translation1"}, ...]
# Enable two-hop translation via pivot
PIVOT_FALLBACK=1
# Pivot language (usually English)
PIVOT_LANG=en
# Log level
LOG_LEVEL=INFO
# Per-request logging (verbose)
REQUEST_LOG=1
# Format
LOG_FORMAT=plain # Human-readable
LOG_FORMAT=json # Structured JSON
# File logging with rotation
LOG_TO_FILE=1
LOG_FILE_PATH=/var/log/marian-translator/app.log
LOG_FILE_MAX_BYTES=10485760 # 10MB
LOG_FILE_BACKUP_COUNT=5
# Include raw text in logs (privacy risk!)
LOG_INCLUDE_TEXT=0
# Periodically clear CUDA cache (seconds, 0=disabled)
CUDA_CACHE_CLEAR_INTERVAL_SEC=0
# Worker count (use 1 for single GPU)
WEB_CONCURRENCY=1
# Request timeout
TIMEOUT=60
# Graceful shutdown timeout
GRACEFUL_TIMEOUT=20
# Keep-alive timeout
KEEP_ALIVE=5
# GET request
curl "http://localhost:8000/translate?target_lang=de&text=Hello%20world&source_lang=en"
# Response
{
"translations": ["Hallo Welt"]
}
# POST request
curl -X POST http://localhost:8000/translate \
-H 'Content-Type: application/json' \
-d '{
"text": [
"Hello world",
"This is a test",
"Machine translation is amazing"
],
"target_lang": "de",
"source_lang": "en",
"beam_size": 1,
"perform_sentence_splitting": true
}'
# Response
{
"target_lang": "de",
"source_lang": "en",
"translated": [
"Hallo Welt",
"Das ist ein Test",
"Maschinenübersetzung ist erstaunlich"
],
"translation_time": 0.342
}
# Omit source_lang for auto-detection
curl -X POST http://localhost:8000/translate \
-H 'Content-Type: application/json' \
-d '{
"text": ["Bonjour le monde"],
"target_lang": "en"
}'
# Response
{
"target_lang": "en",
"source_lang": "fr", # Detected
"translated": ["Hello world"],
"translation_time": 0.156
}
# GET
curl "http://localhost:8000/language_detection?text=Hola%20mundo"
# {"language": "es"}
# POST with batch
curl -X POST http://localhost:8000/language_detection \
-H 'Content-Type: application/json' \
-d '{"text": ["Hello", "Bonjour", "Hola"]}'
# {"languages": ["en", "fr", "es"]}
# Health check
curl http://localhost:8000/healthz
# {"status": "ok"}
# Readiness
curl http://localhost:8000/readyz
# {
# "status": "ready",
# "device": "cuda:0",
# "queue_enabled": true,
# "max_inflight": 1
# }
# Cache status
curl http://localhost:8000/cache
# {
# "capacity": 6,
# "size": 3,
# "keys": ["en->de", "de->en", "fr->en"],
# "device": "cuda:0",
# "inflight": 1,
# "queue_enabled": true
# }
# Model info
curl http://localhost:8000/model_name | jq
# When queue is full, you get 429
curl -X POST http://localhost:8000/translate \
-H 'Content-Type: application/json' \
-d '{"text": ["test"], "target_lang": "de"}'
# Response: 429 Too Many Requests
# Headers: Retry-After: 45
# Body:
{
"message": "Too many requests; queue full",
"retry_after_sec": 45
}
# Proper client behavior:
# 1. Read Retry-After header
# 2. Wait that long + jitter
# 3. Retry request
Gladde retry-after schatting die zich aanpast aan de werkelijke aanvraagduur:
Voorbeeld:
.\build-all.ps1
Wat gebeurt hier?
chmod +x build-all.sh
./build-all.sh
: Zoals een gewogen gemiddelde dat meer belang geeft aan recente waardenSmoothing factor (α):
latest, min, gpu, gpu-minWaarom niet gewoon gemiddeld?20250108.143022: Geeft klanten realistischtijden die zich aanpassen aan de huidige systeembelasting
# Always get the latest version
docker pull scottgal/mostlylucid-nmt:cpu
# Or use the :latest alias
docker pull scottgal/mostlylucid-nmt:latest
# Pin to a specific version for reproducibility
docker pull scottgal/mostlylucid-nmt:cpu-20250108.143022
docker pull scottgal/mostlylucid-nmt:cpu-min-20250108.143022
Middelenbeheer
: Slimme caching, chunking en parallelle verwerking
docker inspect scottgal/mostlylucid-nmt:cpu | jq '.[0].Config.Labels'
Waarneembaarheid: Gedetailleerde logging en metrics tracking.
# Using pre-built image from Docker Hub (recommended)
docker run -d \
--name translator \
-p 8000:8000 \
-e ENABLE_QUEUE=1 \
-e MAX_QUEUE_SIZE=500 \
-e EASYNMT_BATCH_SIZE=16 \
-e TIMEOUT=180 \
-e LOG_LEVEL=INFO \
-e REQUEST_LOG=0 \
scottgal/mostlylucid-nmt
# Or build locally
docker build -t mostlylucid-nmt .
docker run -d --name translator -p 8000:8000 mostlylucid-nmt
# Check logs
docker logs -f translator
# Using pre-built GPU image from Docker Hub (recommended)
docker run -d \
--name translator-gpu \
--gpus all \
-p 8000:8000 \
-e USE_GPU=true \
-e DEVICE=cuda:0 \
-e PRELOAD_MODELS="en->de,de->en,en->fr,fr->en,en->es,es->en" \
-e EASYNMT_MODEL_ARGS='{"torch_dtype":"fp16"}' \
-e EASYNMT_BATCH_SIZE=64 \
-e MAX_CACHED_MODELS=8 \
-e ENABLE_QUEUE=1 \
-e MAX_QUEUE_SIZE=2000 \
-e WEB_CONCURRENCY=1 \
-e TIMEOUT=180 \
-e GRACEFUL_TIMEOUT=30 \
-e LOG_FORMAT=json \
-e LOG_TO_FILE=1 \
-v /var/log/translator:/var/log/marian-translator \
scottgal/mostlylucid-nmt:gpu
# Or build locally
docker build -f Dockerfile.gpu -t mostlylucid-nmt:gpu .
docker run -d --name translator-gpu --gpus all -p 8000:8000 mostlylucid-nmt:gpu
# Monitor cache and performance
watch -n 5 "curl -s http://localhost:8000/cache | jq"
version: '3.8'
services:
translator:
image: scottgal/mostlylucid-nmt:gpu # Use pre-built image
container_name: translator
restart: unless-stopped
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: 1
capabilities: [gpu]
ports:
- "8000:8000"
environment:
USE_GPU: "true"
DEVICE: "cuda:0"
PRELOAD_MODELS: "en->de,de->en,en->fr,fr->en"
EASYNMT_MODEL_ARGS: '{"torch_dtype":"fp16"}'
EASYNMT_BATCH_SIZE: "64"
MAX_CACHED_MODELS: "8"
ENABLE_QUEUE: "1"
MAX_QUEUE_SIZE: "2000"
WEB_CONCURRENCY: "1"
TIMEOUT: "180"
LOG_FORMAT: "json"
LOG_TO_FILE: "1"
volumes:
- translator-logs:/var/log/marian-translator
- translator-cache:/root/.cache/huggingface
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:8000/healthz"]
interval: 30s
timeout: 10s
retries: 3
start_period: 40s
volumes:
translator-logs:
translator-cache:
apiVersion: apps/v1
kind: Deployment
metadata:
name: translator
spec:
replicas: 2 # Scale horizontally for CPU, use 1 per GPU
selector:
matchLabels:
app: translator
template:
metadata:
labels:
app: translator
spec:
containers:
- name: translator
image: scottgal/mostlylucid-nmt:gpu
ports:
- containerPort: 8000
env:
- name: USE_GPU
value: "true"
- name: EASYNMT_MODEL_ARGS
value: '{"torch_dtype":"fp16"}'
- name: PRELOAD_MODELS
value: "en->de,de->en"
- name: ENABLE_QUEUE
value: "1"
- name: MAX_QUEUE_SIZE
value: "2000"
resources:
requests:
memory: "4Gi"
cpu: "2"
nvidia.com/gpu: 1
limits:
memory: "8Gi"
cpu: "4"
nvidia.com/gpu: 1
livenessProbe:
httpGet:
path: /healthz
port: 8000
initialDelaySeconds: 30
periodSeconds: 10
readinessProbe:
httpGet:
path: /readyz
port: 8000
initialDelaySeconds: 20
periodSeconds: 5
---
apiVersion: v1
kind: Service
metadata:
name: translator
spec:
selector:
app: translator
ports:
- port: 80
targetPort: 8000
type: LoadBalancer
MODEL_FAMILLY
EASYNMT_MODEL_ARGS='{"torch_dtype":"fp16"}'
: Goede kwaliteit, 100 talen, enkele 2.2GB model
# Start high, reduce if you get OOM
EASYNMT_BATCH_SIZE=64 # Try 128 on large GPUs
AUTO_MODEL_FALLBACK
PRELOAD_MODELS="en->de,de->en,en->fr,fr->en,en->es,es->en"
: Probeer automatisch andere families als het paar niet beschikbaar is
WEB_CONCURRENCY=1
MAX_INFLIGHT_TRANSLATIONS=1
(standaard): ingeschakeld - maximale dekking
MAX_CACHED_MODELS=10 # Keep more models in VRAM
: Gehandicapt - strikte eengezinsmodus
# beam_size=1 is 3-5x faster than beam_size=5
# Quality difference is often minimal
curl -X POST ... -d '{"beam_size": 1, ...}'
: Prioriteitsorde voor terugval
EASYNMT_BATCH_SIZE=8
Standaard:
MAX_WORKERS_BACKEND=4
MAX_INFLIGHT_TRANSLATIONS=4
WEB_CONCURRENCY=2
(kwaliteit eerst)
PERFORM_SENTENCE_SPLITTING_DEFAULT=0
(dekking eerst)
// Bad: 100 separate requests
for (const text of texts) {
await translate(text);
}
// Good: 1 batch request
await translate(texts);
MODEL_CACHE_DIR
async function translateWithRetry(texts) {
try {
return await translate(texts);
} catch (err) {
if (err.status === 429) {
const retryAfter = err.headers['retry-after'];
const jitter = Math.random() * 5;
await sleep((retryAfter + jitter) * 1000);
return translateWithRetry(texts);
}
throw err;
}
}
: Persistente modelopslag via Docker volumes
// Reuse HTTP connections
const agent = new https.Agent({ keepAlive: true });
Instellen op
// Bad: mixed language pairs in one request
translate([
{ text: "Hello", sourceLang: "en", targetLang: "de" },
{ text: "Bonjour", sourceLang: "fr", targetLang: "de" }
]);
// Good: group by language pair
translateBatch(enToDe, "en", "de");
translateBatch(frToDe, "fr", "de");
Voorbeelden van gebruik
translation_requests_total{lang_pair="en->de",status="success"} 1523
translation_requests_total{lang_pair="en->de",status="error"} 7
translation_duration_seconds{lang_pair="en->de",quantile="0.5"} 0.342
translation_duration_seconds{lang_pair="en->de",quantile="0.95"} 1.234
translation_queue_depth 23
translation_cache_size 6
translation_cache_hits_total 8234
translation_cache_misses_total 142
# Enable JSON logging
LOG_FORMAT=json REQUEST_LOG=1
# Output example
{
"ts": "2025-01-08T15:30:45+0000",
"level": "INFO",
"name": "app",
"message": "translate_post done items=5 dt=0.342s",
"req_id": "a3d2f5b1-c4e6-4f7a-9d8c-1e2f3a4b5c6d",
"endpoint": "/translate",
"src": "en",
"tgt": "de",
"items": 5,
"duration_ms": 342
}
Batchvertaling (aanbevolen)
Alleen taaldetectie |---------|---------|-----------------| | ObservabiliteitseindpuntenOmgaan met Backpressure | Bouwen en versierenAlle Docker-afbeeldingen bevatten nu de juiste versiering en metadata voor het volgen. | Snel bouwenBouw alle 4 varianten met automatische datetime versiering: | **Vensters:**Linux/Mac: | VersiestrategieElk gebouw creëert | twee tagsBenoemde tag | ) - wijst altijd op de meest recenteVersie-tag | (b.v.,) - onveranderlijke snapshot | **Bijvoorbeeld:**OCI-etiketten | **Elke afbeelding bevat metadata:**Versie | **: Bouwtijdstempel (JJJJMMDD.HHMMSS)**Bouwdatum | : ISO 8601 tijdstempelGit commit
: cpu-full, cpu-min, gpu-full, of gpu-minInspecteer labels:
Voor gedetailleerde bouwinstructies en CI/CD-integratie, zie
MAX_QUEUE_SIZEMAX_INFLIGHT_TRANSLATIONSGPU-implementatieOptimalisatie van de prestatiesGPU Optimalisatie Checklist
FP16-precisie gebruiken
ENABLE_QUEUE=1Groepsgrootte instellenHete modellen voorladen
Enige werknemer per GPU
EASYNMT_BATCH_SIZEMAX_CACHED_MODELSEASYNMT_MODEL_ARGS='{"torch_dtype":"fp16"}'WEB_CONCURRENCY=1Parallellisme verhogenMAX_INFLIGHT_TRANSLATIONS=1Best practices van opdrachtgeversPartijverzoeken
Respect opnieuw proberen-na
PRELOAD_MODELS="en->de,de->en"
Groeperen op taalpaar Helsinki-NLP/opus-mt-{src}-{tgt}Monitoring en Waarneming
Sleutel Metrics naar Track
PIVOT_FALLBACK=1(verzoeken/sec)curl http://localhost:8000/lang_pairsWachtrijdiepte(huidige wachttijd)
Cache hit rate
MASK_EMOJI=0FoutpercentageMASK_PUNCT=0SYMBOL_MASKING=0(indien van toepassing)
public class MostlyLucidNmtClient
{
private readonly HttpClient _httpClient;
private readonly string _baseUrl;
public MostlyLucidNmtClient(HttpClient httpClient, string baseUrl)
{
_httpClient = httpClient;
_baseUrl = baseUrl;
}
public async Task<TranslationResponse> TranslateAsync(
List<string> texts,
string targetLang,
string sourceLang = "",
int beamSize = 1,
bool performSentenceSplitting = true,
CancellationToken cancellationToken = default)
{
var request = new TranslationRequest
{
Text = texts,
TargetLang = targetLang,
SourceLang = sourceLang,
BeamSize = beamSize,
PerformSentenceSplitting = performSentenceSplitting
};
var response = await _httpClient.PostAsJsonAsync(
$"{_baseUrl}/translate",
request,
cancellationToken);
if (response.StatusCode == System.Net.HttpStatusCode.TooManyRequests)
{
// Read Retry-After header
var retryAfter = response.Headers.RetryAfter?.Delta?.TotalSeconds ?? 30;
var jitter = Random.Shared.Next(0, 5);
await Task.Delay(TimeSpan.FromSeconds(retryAfter + jitter), cancellationToken);
// Retry
return await TranslateAsync(texts, targetLang, sourceLang, beamSize,
performSentenceSplitting, cancellationToken);
}
response.EnsureSuccessStatusCode();
return await response.Content.ReadFromJsonAsync<TranslationResponse>(cancellationToken);
}
}
public class TranslationRequest
{
[JsonPropertyName("text")]
public List<string> Text { get; set; }
[JsonPropertyName("target_lang")]
public string TargetLang { get; set; }
[JsonPropertyName("source_lang")]
public string SourceLang { get; set; }
[JsonPropertyName("beam_size")]
public int BeamSize { get; set; }
[JsonPropertyName("perform_sentence_splitting")]
public bool PerformSentenceSplitting { get; set; }
}
public class TranslationResponse
{
[JsonPropertyName("target_lang")]
public string TargetLang { get; set; }
[JsonPropertyName("source_lang")]
public string SourceLang { get; set; }
[JsonPropertyName("translated")]
public List<string> Translated { get; set; }
[JsonPropertyName("translation_time")]
public double TranslationTime { get; set; }
}
Geheugengebruik
services.AddHttpClient<MostlyLucidNmtClient>(client =>
{
client.BaseAddress = new Uri("http://translator:8000");
client.Timeout = TimeSpan.FromMinutes(3);
});
Voorbeeld Prometheus Metrics**Als u Prometheus integreert (niet ingebouwd, maar eenvoudig toe te voegen):**Gestructureerd logging-voorbeeld
Vergelijking: EasyNMT vs. MeestalLucid-NMT
Invoerafhandeling
Waarneembaarheid
Handmatig, geen caching LRU cache met auto-eviction
Configuratie40+ env vars voor fine-tuningAPI-compatibiliteit
# Maximum coverage with auto-fallback (recommended!)
docker run -d -p 8000:8000 \
-v ./model-cache:/models \
-e MODEL_CACHE_DIR=/models \
-e AUTO_MODEL_FALLBACK=1 \
-e MODEL_FALLBACK_ORDER="opus-mt,mbart50,m2m100" \
scottgal/mostlylucid-nmt:cpu-min
# GPU with best quality
docker run -d --gpus all -p 8000:8000 \
-e USE_GPU=true \
-e MODEL_FAMILY=opus-mt \
-e EASYNMT_MODEL_ARGS='{"torch_dtype":"fp16"}' \
scottgal/mostlylucid-nmt:gpu
# Test it
curl -X POST http://localhost:8000/translate \
-H 'Content-Type: application/json' \
-d '{"text": ["Hello world"], "target_lang": "de"}'
OOM (Out of Memory) op GPU
Oorzaak:
Batch grootte te hoog of te veel modellen gecached.
Traag eerste verzoek[Oorzaak:
Translation NMT Neural Machine Translation Python FastAPI Docker CUDA PyTorch Transformers Helsinki-NLP Production Microservices API
© 2026 Scott Galloway — Unlicense — All content and source code on this site is free to use, copy, modify, and sell.