laya 0.3.2__tar.gz → 0.3.3__tar.gz

{laya-0.3.2 → laya-0.3.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: laya
-Version: 0.3.2
+Version: 0.3.3
 Summary: Fast, non-autoregressive System 1 decision engine with calibrated probabilities
 Author: Convai Innovations
 License: Apache-2.0
@@ -246,10 +246,39 @@ Precedence, highest first:
 5. Detected script (exact) and, for Latin text, a stopword/diacritic language guess (best effort).
 6. `default=` (`"english"` unless you change it).
 
+### Preload — make routing free
+
+A cold checkpoint build costs **seconds**; language detection costs **microseconds**. At the
+default `max_loaded=1`, traffic that alternates languages rebuilds a model on *every* request.
+For a server or a demo, preload:
+
+```python
+router = Router(preload=True)                  # every checkpoint resident, routing is free
+router = Router(preload=True, device="cuda")
+router.preload(["english", "multilingual"])    # or just the two you serve
+```
+
+`preload` raises `max_loaded` to fit what it built, so the LRU cannot evict it immediately.
+
+If the process already has a checkpoint loaded for other reasons, hand it over instead of
+loading a second copy:
+
+```python
+router.attach("english", existing_agent)   # no duplicate 421M parameters
+router.preload()                           # builds only what is still missing
+```
+
+Measured on CPU with the demo Space's own workload:
+
+| | per request | model loads |
+|---|---|---|
+| `Router()` — lazy, `max_loaded=1` | 4–6 s on every language switch | 1 per switch |
+| `Router(preload=True)` | **193–464 ms** | **none** |
+
 ### Memory
 
-All three together are ~1.16B parameters, so `Router` keeps **one** resident by default and
-evicts least-recently-used:
+All three together are ~1.16B parameters (~4.6 GB fp32), so `Router` keeps **one** resident by
+default and evicts least-recently-used. Raise it when you have the RAM:
 
 ```python
 Router(max_loaded=2)       # keep two hot

{laya-0.3.2 → laya-0.3.3}/README.md

@@ -216,10 +216,39 @@ Precedence, highest first:
 5. Detected script (exact) and, for Latin text, a stopword/diacritic language guess (best effort).
 6. `default=` (`"english"` unless you change it).
 
+### Preload — make routing free
+
+A cold checkpoint build costs **seconds**; language detection costs **microseconds**. At the
+default `max_loaded=1`, traffic that alternates languages rebuilds a model on *every* request.
+For a server or a demo, preload:
+
+```python
+router = Router(preload=True)                  # every checkpoint resident, routing is free
+router = Router(preload=True, device="cuda")
+router.preload(["english", "multilingual"])    # or just the two you serve
+```
+
+`preload` raises `max_loaded` to fit what it built, so the LRU cannot evict it immediately.
+
+If the process already has a checkpoint loaded for other reasons, hand it over instead of
+loading a second copy:
+
+```python
+router.attach("english", existing_agent)   # no duplicate 421M parameters
+router.preload()                           # builds only what is still missing
+```
+
+Measured on CPU with the demo Space's own workload:
+
+| | per request | model loads |
+|---|---|---|
+| `Router()` — lazy, `max_loaded=1` | 4–6 s on every language switch | 1 per switch |
+| `Router(preload=True)` | **193–464 ms** | **none** |
+
 ### Memory
 
-All three together are ~1.16B parameters, so `Router` keeps **one** resident by default and
-evicts least-recently-used:
+All three together are ~1.16B parameters (~4.6 GB fp32), so `Router` keeps **one** resident by
+default and evicts least-recently-used. Raise it when you have the RAM:
 
 ```python
 Router(max_loaded=2)       # keep two hot

{laya-0.3.2 → laya-0.3.3}/laya/__init__.py

@@ -22,7 +22,7 @@ from .presets import (
 )
 from .router import DEFAULT_MODELS, RouteDecision, Router
 
-__version__ = "0.3.2"
+__version__ = "0.3.3"
 __all__ = [
     "Agent",
     "RLAgent",

{laya-0.3.2 → laya-0.3.3}/laya/router.py

@@ -194,6 +194,19 @@ class Router:
                 if k not in self._order:
                     self._agents.pop(k, None)
 
+    def attach(self, name: str, agent: Any):
+        """Register an already-built Agent under `name` instead of loading a second copy.
+
+        Useful when the process has a checkpoint loaded for other reasons: a demo that already
+        built `convaiinnovations/laya` can hand it to the router rather than pay for -- and hold
+        in memory -- a duplicate 421M parameters.
+        """
+        key = normalise_name(name)
+        self._agents[key] = agent
+        self._touch(key)
+        self.max_loaded = max(self.max_loaded, len(self._agents))
+        return agent
+
     def preload(self, names: Optional[List[str]] = None):
         """Download and build checkpoints up front so no request ever pays a model load.
 
@@ -203,9 +216,10 @@ class Router:
         the LRU would immediately evict what this just built.
         """
         names = [normalise_name(n) for n in (names or list(self.models))]
-        self.max_loaded = max(self.max_loaded, len(names))
+        self.max_loaded = max(self.max_loaded, len(names), len(self._agents))
         for n in names:
-            self.load(n)
+            if n not in self._agents:      # an attached agent is already built
+                self.load(n)
         return self
 
     def unload(self, name: Optional[str] = None):

{laya-0.3.2 → laya-0.3.3}/laya.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: laya
-Version: 0.3.2
+Version: 0.3.3
 Summary: Fast, non-autoregressive System 1 decision engine with calibrated probabilities
 Author: Convai Innovations
 License: Apache-2.0
@@ -246,10 +246,39 @@ Precedence, highest first:
 5. Detected script (exact) and, for Latin text, a stopword/diacritic language guess (best effort).
 6. `default=` (`"english"` unless you change it).
 
+### Preload — make routing free
+
+A cold checkpoint build costs **seconds**; language detection costs **microseconds**. At the
+default `max_loaded=1`, traffic that alternates languages rebuilds a model on *every* request.
+For a server or a demo, preload:
+
+```python
+router = Router(preload=True)                  # every checkpoint resident, routing is free
+router = Router(preload=True, device="cuda")
+router.preload(["english", "multilingual"])    # or just the two you serve
+```
+
+`preload` raises `max_loaded` to fit what it built, so the LRU cannot evict it immediately.
+
+If the process already has a checkpoint loaded for other reasons, hand it over instead of
+loading a second copy:
+
+```python
+router.attach("english", existing_agent)   # no duplicate 421M parameters
+router.preload()                           # builds only what is still missing
+```
+
+Measured on CPU with the demo Space's own workload:
+
+| | per request | model loads |
+|---|---|---|
+| `Router()` — lazy, `max_loaded=1` | 4–6 s on every language switch | 1 per switch |
+| `Router(preload=True)` | **193–464 ms** | **none** |
+
 ### Memory
 
-All three together are ~1.16B parameters, so `Router` keeps **one** resident by default and
-evicts least-recently-used:
+All three together are ~1.16B parameters (~4.6 GB fp32), so `Router` keeps **one** resident by
+default and evicts least-recently-used. Raise it when you have the RAM:
 
 ```python
 Router(max_loaded=2)       # keep two hot

{laya-0.3.2 → laya-0.3.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "laya"
-version = "0.3.2"
+version = "0.3.3"
 description = "Fast, non-autoregressive System 1 decision engine with calibrated probabilities"
 readme = "README.md"
 requires-python = ">=3.8"

{laya-0.3.2 → laya-0.3.3}/setup.py

@@ -2,7 +2,7 @@ from setuptools import setup, find_packages
 
 setup(
     name="laya",
-    version="0.3.2",
+    version="0.3.3",
     description="Fast, non-autoregressive System 1 decision engine with calibrated probabilities",
     long_description=open("README.md").read(),
     long_description_content_type="text/markdown",

{laya-0.3.2 → laya-0.3.3}/tests/test_router.py

@@ -255,6 +255,21 @@ _load_stub(rp2, "english")
 check("preload/touch does not evict", sorted(rp2.loaded), ["english", "multilingual"])
 
 
+# --------------------------------------------------------------------- attach
+ra = stubbed_router(1)
+sentinel = _Stub("already-built")
+ra.attach("english", sentinel)
+check("attach/registers under the name", ra._agents["english"], sentinel)
+check("attach/counts as resident", "english" in ra.loaded, True)
+check("attach/raises max_loaded to hold it", ra.max_loaded >= 1, True)
+# attaching then loading another must not evict the attached one
+ra.max_loaded = max(ra.max_loaded, 2)
+_load_stub(ra, "multilingual")
+check("attach/survives a later load", sorted(ra.loaded), ["english", "multilingual"])
+check("attach/still the same object", ra._agents["english"] is sentinel, True)
+check("attach/accepts aliases", stubbed_router(1).attach("en", _Stub("x")) is not None, True)
+
+
 # --------------------------------------------------------------------- report
 print("\n%d passed, %d failed" % (len(PASS), len(FAIL)))
 for f in FAIL: