[refactor] rename package mini_fastvideo → v2

Directory rename (git mv, history preserved) plus rewrite of all references: absolute imports in
tests, the zero-dep runner, docstrings, comments, and the README. No behavior change.

Run: python3 -m pytest v2/tests/ -q ; python3 v2/run_tests.py ; python3 -m v2.examples

v2/README.md

@@ -1,4 +1,4 @@
-# mini-fastvideo
+# v2
 
 A scoped, **CPU-testable** realization of the `design_v3.md` model-native runtime — the architecture
 where *omni is native, train and serve are the same loops by construction, and correctness is a typed
@@ -76,9 +76,9 @@ engine never imports `training`. `card/` imports no runtime.
 
 ```bash
 cd /Users/willlin/src/FastVideo-mini
-python3 -m pytest mini_fastvideo/tests/ -q      # 49 tests
-python3 mini_fastvideo/run_tests.py             # same suite, ZERO deps (no pytest needed)
-python3 -m mini_fastvideo.examples              # the design_v3 §15 worked examples
+python3 -m pytest v2/tests/ -q      # 49 tests
+python3 v2/run_tests.py             # same suite, ZERO deps (no pytest needed)
+python3 -m v2.examples              # the design_v3 §15 worked examples
 ```
 
 Requires only Python 3.10+ and numpy.
@@ -111,7 +111,7 @@ DAG-of-engines can express, made native:
 The diffusion loop is literally `WanDenoiseLoop` (the same step body the engine serves for Wan), bound
 to the MoT module — one loop definition, reused. The interleave gate generalizes across loop types
 (`test_omni.py::test_omni_interleave_parity_holds_across_loop_types`). Run it:
-`python3 -c "from mini_fastvideo.models import build_omni_engine; ..."` or see `mini_fastvideo/examples.py`.
+`python3 -c "from v2.models import build_omni_engine; ..."` or see `v2/examples.py`.
 
 Still future work (declared on the spine, not yet built): the packed factored-sequence
 (`[text|vision|action|sound]`) `LoopState.extension`, joint multi-modality denoise with per-modality
@@ -144,9 +144,9 @@ The serving layer is built, not deferred to Dynamo:
 
 ```python
 import asyncio
-from mini_fastvideo.models import build_default_engine, build_omni_engine
-from mini_fastvideo.runtime import AsyncEngine
-from mini_fastvideo.serving import OmniOpenAIServer
+from v2.models import build_default_engine, build_omni_engine
+from v2.runtime import AsyncEngine
+from v2.serving import OmniOpenAIServer
 async def serve():
     eng = build_default_engine(); build_omni_engine(eng)
     host, port = await OmniOpenAIServer(AsyncEngine(eng)).serve(port=8000)

v2/__init__.py

@@ -1,4 +1,4 @@
-"""mini-fastvideo — a scoped, CPU-testable realization of the design_v3 model-native runtime.
+"""v2 — a scoped, CPU-testable realization of the design_v3 model-native runtime.
 
 > A model card is a (recipe, runtime) pair with a parity obligation.
 > The model owns loop semantics; the runtime owns loop lifecycle.
@@ -12,7 +12,7 @@ omni-ready (multi-loop ModelInstance, ar_decode/chunk_rollout loop kinds) for th
 Cosmos3 + vllm-omni omni ports.
 
 The core is numpy-only and CPU-testable; heavy Wan/LTX neural forwards become lazy torch
-adapters (see ``mini_fastvideo/models/*/components.py``) that are off the test path.
+adapters (see ``v2/models/*/components.py``) that are off the test path.
 """
 from __future__ import annotations
 

v2/_types.py

@@ -1,10 +1,10 @@
-"""Shared lightweight type aliases for mini-fastvideo.
+"""Shared lightweight type aliases for v2.
 
 The core (card/loop/runtime/cache/parity/program/request/training) is numpy-only
 and CPU-testable. Tensors are therefore typed structurally as ``TensorLike``:
 on the CPU test path they are ``numpy.ndarray``; on a GPU box the same code runs
 with ``torch.Tensor``. The core never imports torch — only model-component
-*adapters* do, lazily (see ``mini_fastvideo/card/components.py``).
+*adapters* do, lazily (see ``v2/card/components.py``).
 """
 from __future__ import annotations
 

v2/card/specs.py

@@ -143,7 +143,7 @@ class ComponentSpec:
     resident_for: list[str] = field(default_factory=list)  # loop_ids that keep this resident mid-request
     optional_for: set[str] = field(default_factory=set)    # tasks that don't need it
     required_for: set[str] = field(default_factory=set)    # tasks that require it
-    # mini-fastvideo wiring: a factory producing the live component (toy numpy or torch adapter)
+    # v2 wiring: a factory producing the live component (toy numpy or torch adapter)
     factory: Callable[..., Any] | None = None
 
 
@@ -170,7 +170,7 @@ class LoopSpec:
     # omni-ready (omni-readiness report §4a):
     shared_weight_components: list[str] = field(default_factory=list)
     allows_interleaving: bool = True
-    # mini-fastvideo: the Loop implementation factory (built at bind time)
+    # v2: the Loop implementation factory (built at bind time)
     loop_factory: Callable[..., Any] | None = None
 
 

v2/examples.py

@@ -1,6 +1,6 @@
 """Worked examples (design_v3 §15), runnable as a demo:
 
-    python3 -m mini_fastvideo.examples
+    python3 -m v2.examples
 
 Each prints what it demonstrates. This doubles as living documentation of the API.
 """
@@ -136,7 +136,7 @@ def example_h_serving_and_fleet() -> None:
 
 def main() -> None:
     print("=" * 78)
-    print("mini-fastvideo — worked examples (design_v3 §6,§13-16). One runtime, many loops.")
+    print("v2 — worked examples (design_v3 §6,§13-16). One runtime, many loops.")
     print("=" * 78)
     eng = build_default_engine()
     print(f"registered (recipe, runtime) cards: {list(eng._registry)}")

v2/request/streams.py

@@ -6,7 +6,7 @@ logs, finals." Event taxonomy: ``request.*``, ``session.*``, ``artifact.*``,
 stream, byte-range/buffer-ref, codec/container, timestamp range, and
 preview-vs-final — invalid combinations are unrepresentable.
 
-For mini-fastvideo the Stream is a simple synchronous ordered channel (the core is
+For v2 the Stream is a simple synchronous ordered channel (the core is
 numpy/sync and CPU-testable); a GPU/server deployment swaps in an async channel
 without changing the event types.
 """

v2/run_tests.py

@@ -1,13 +1,13 @@
 #!/usr/bin/env python3
-"""Zero-dependency test runner for mini-fastvideo.
+"""Zero-dependency test runner for v2.
 
-The mini-fastvideo core is numpy-only and CPU-testable. This runner discovers
-``test_*`` functions in ``mini_fastvideo/tests/test_*.py`` and runs them without
+The v2 core is numpy-only and CPU-testable. This runner discovers
+``test_*`` functions in ``v2/tests/test_*.py`` and runs them without
 needing pytest installed (it is fully pytest-compatible, so ``pytest`` works too).
 
 Usage:
-    python3 mini_fastvideo/run_tests.py            # run all
-    python3 mini_fastvideo/run_tests.py interleave # run files matching a substring
+    python3 v2/run_tests.py            # run all
+    python3 v2/run_tests.py interleave # run files matching a substring
 """
 from __future__ import annotations
 
@@ -18,7 +18,7 @@ from pathlib import Path
 
 HERE = Path(__file__).resolve().parent
 TESTS_DIR = HERE / "tests"
-REPO_ROOT = HERE.parent  # so `import mini_fastvideo` works
+REPO_ROOT = HERE.parent  # so `import v2` works
 
 
 def _load(path: Path):

v2/serving/server.py

@@ -61,7 +61,7 @@ def _ser_artifacts(out: Output) -> dict:
 
 
 class OmniOpenAIServer:
-    def __init__(self, engine: Any, *, engine_id: str = "mini-fastvideo", max_jobs: int = 512):
+    def __init__(self, engine: Any, *, engine_id: str = "v2", max_jobs: int = 512):
         self.engine = engine                       # an AsyncEngine
         self.engine_id = engine_id
         self.jobs: dict[str, dict] = {}            # video job store

v2/tests/test_admission.py

@@ -7,12 +7,12 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo.memory import MemoryManager, OutOfMemory
-from mini_fastvideo.models import build_default_engine
-from mini_fastvideo.models.wan21.loop import latent_shape
-from mini_fastvideo.parity import compare_outputs
-from mini_fastvideo.request import DiffusionParams, TaskType, make_request
-from mini_fastvideo.runtime import AdmissionController, Engine
+from v2.memory import MemoryManager, OutOfMemory
+from v2.models import build_default_engine
+from v2.models.wan21.loop import latent_shape
+from v2.parity import compare_outputs
+from v2.request import DiffusionParams, TaskType, make_request
+from v2.runtime import AdmissionController, Engine
 
 
 def test_memory_manager_reserve_release_and_oom():

v2/tests/test_cache.py

@@ -3,9 +3,9 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo.cache import CacheKey, CacheManager, CachePolicy
-from mini_fastvideo.cache.classes import FeatureCache, Slab, SlabKVCache
-from mini_fastvideo.models.wan_causal import build_wan_causal_card
+from v2.cache import CacheKey, CacheManager, CachePolicy
+from v2.cache.classes import FeatureCache, Slab, SlabKVCache
+from v2.models.wan_causal import build_wan_causal_card
 
 
 def test_feature_cache_reuse_and_adapter_partition():
@@ -55,6 +55,6 @@ def test_cache_manager_materializes_only_declared_classes():
     cm = CacheManager.from_card(build_wan_causal_card())
     assert cm.has("feature") and cm.has("slab_kv")
     # a pure bidirectional card declares no KV class
-    from mini_fastvideo.models.wan21 import build_wan21_card
+    from v2.models.wan21 import build_wan21_card
     cm2 = CacheManager.from_card(build_wan21_card())
     assert cm2.has("feature") and not cm2.has("slab_kv")    # KV is the minority case

v2/tests/test_contracts.py

@@ -3,18 +3,18 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo._enums import ConsistencyLevel, LoopKind, WorkUnitKind
-from mini_fastvideo.card import (
+from v2._enums import ConsistencyLevel, LoopKind, WorkUnitKind
+from v2.card import (
     CardValidationError,
     CostModel,
     LoopSpec,
     ModelCard,
     RecipeSpec,
 )
-from mini_fastvideo.cache.keys import CacheKey, content_hash
-from mini_fastvideo.models.wan21 import build_wan21_card
-from mini_fastvideo.models.wan_causal import build_wan_causal_card
-from mini_fastvideo.parallel import ParallelPlan, ParallelValidationError, validate_plan
+from v2.cache.keys import CacheKey, content_hash
+from v2.models.wan21 import build_wan21_card
+from v2.models.wan_causal import build_wan_causal_card
+from v2.parallel import ParallelPlan, ParallelValidationError, validate_plan
 
 
 def test_wan_card_validates():

v2/tests/test_extensions.py

@@ -4,19 +4,19 @@ interleave-safe; a module-global version provably is NOT — which is *why* §11
 """
 from __future__ import annotations
 
-from mini_fastvideo._enums import WorkUnitKind
-from mini_fastvideo.card import CostModel
-from mini_fastvideo.extend import (
+from v2._enums import WorkUnitKind
+from v2.card import CostModel
+from v2.extend import (
     InterceptorChain,
     InterceptorConflict,
     NaNWatch,
     Profiler,
     ResidualSkipInterceptor,
 )
-from mini_fastvideo.models import build_default_engine
-from mini_fastvideo.parity import assert_interleave_parity
-from mini_fastvideo.request import DiffusionParams, TaskType, make_request
-from mini_fastvideo.runtime import Engine
+from v2.models import build_default_engine
+from v2.parity import assert_interleave_parity
+from v2.request import DiffusionParams, TaskType, make_request
+from v2.runtime import Engine
 
 
 def _req(prompt, seed):

v2/tests/test_inference.py

@@ -3,11 +3,11 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo.models import build_default_engine
-from mini_fastvideo.models.ltx2 import BASE_SIGMAS, REFINE_SIGMAS, build_ltx2_card
-from mini_fastvideo.models.wan21 import build_wan21_card
-from mini_fastvideo.models.wan_causal import build_wan_causal_card
-from mini_fastvideo.request import DiffusionParams, OutputSpec, TaskType, make_request
+from v2.models import build_default_engine
+from v2.models.ltx2 import BASE_SIGMAS, REFINE_SIGMAS, build_ltx2_card
+from v2.models.wan21 import build_wan21_card
+from v2.models.wan_causal import build_wan_causal_card
+from v2.request import DiffusionParams, OutputSpec, TaskType, make_request
 
 
 def _eng():

v2/tests/test_interleave_gate.py

@@ -6,9 +6,9 @@ this gate is not.
 """
 from __future__ import annotations
 
-from mini_fastvideo.models import build_default_engine
-from mini_fastvideo.parity import assert_interleave_parity
-from mini_fastvideo.request import DiffusionParams, TaskType, make_request
+from v2.models import build_default_engine
+from v2.parity import assert_interleave_parity
+from v2.request import DiffusionParams, TaskType, make_request
 
 
 def _req(mid, prompt, seed, steps=4):

v2/tests/test_omni.py

@@ -8,11 +8,11 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo._enums import WorkUnitKind
-from mini_fastvideo.models import build_omni_engine
-from mini_fastvideo.models.cosmos3 import build_cosmos3_card
-from mini_fastvideo.parity import assert_interleave_parity
-from mini_fastvideo.request import DiffusionParams, SamplingParams, TaskType, make_request
+from v2._enums import WorkUnitKind
+from v2.models import build_omni_engine
+from v2.models.cosmos3 import build_cosmos3_card
+from v2.parity import assert_interleave_parity
+from v2.request import DiffusionParams, SamplingParams, TaskType, make_request
 
 
 def _eng():

v2/tests/test_parity.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo.parity import ConsistencyLevel, ParityAligner, array_diff, bit_identical, within
+from v2.parity import ConsistencyLevel, ParityAligner, array_diff, bit_identical, within
 
 
 def test_aligner_no_divergence_on_identical_runs():

v2/tests/test_policies.py

@@ -3,8 +3,8 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo.loop.contracts import StepContext
-from mini_fastvideo.loop.policies import (
+from v2.loop.contracts import StepContext
+from v2.loop.policies import (
     AdaptiveGateCFG,
     BoundaryTimestepRouting,
     ClassicCFG,

v2/tests/test_review_fixes.py

@@ -9,17 +9,17 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo._enums import WorkUnitKind
-from mini_fastvideo.cache import CacheManager
-from mini_fastvideo.card import load_card
-from mini_fastvideo.loop.contracts import ResourceRequest, ShapeSignature, WorkPlan
-from mini_fastvideo.memory import MemoryManager
-from mini_fastvideo.models import build_default_engine
-from mini_fastvideo.models.common import cached_text_encode
-from mini_fastvideo.models.wan21 import build_wan21_card
-from mini_fastvideo.parity import compare_outputs
-from mini_fastvideo.request import DiffusionParams, Output, TaskType, VideoArtifact, make_request
-from mini_fastvideo.runtime import AdmissionController, AdmissionInfeasible, Engine
+from v2._enums import WorkUnitKind
+from v2.cache import CacheManager
+from v2.card import load_card
+from v2.loop.contracts import ResourceRequest, ShapeSignature, WorkPlan
+from v2.memory import MemoryManager
+from v2.models import build_default_engine
+from v2.models.common import cached_text_encode
+from v2.models.wan21 import build_wan21_card
+from v2.parity import compare_outputs
+from v2.request import DiffusionParams, Output, TaskType, VideoArtifact, make_request
+from v2.runtime import AdmissionController, AdmissionInfeasible, Engine
 
 
 def test_infeasible_reservation_fails_fast_not_busy_spin():

v2/tests/test_serving.py

@@ -10,21 +10,21 @@ import asyncio
 
 import numpy as np
 
-from mini_fastvideo._enums import WorkUnitKind
-from mini_fastvideo.card import CostModel
-from mini_fastvideo.deploy import (
+from v2._enums import WorkUnitKind
+from v2.card import CostModel
+from v2.deploy import (
     DynamoWorkerAdapter,
     FakeDynamoRuntime,
     LocalFleet,
     NoWorkerAvailable,
     build_deployment_card,
 )
-from mini_fastvideo.extend import InterceptorChain, ObserverBus
-from mini_fastvideo.models import build_default_engine, build_omni_engine, build_wan21_card, build_wan_t2v_program
-from mini_fastvideo.request import DiffusionParams, OutputSpec, SamplingParams, TaskType, make_request
-from mini_fastvideo.runtime import AsyncEngine, DisaggregatedRunner, Engine, PoolSet, wan_t2v_disaggregated
-from mini_fastvideo.serving import OmniOpenAIServer
-from mini_fastvideo.transport import (
+from v2.extend import InterceptorChain, ObserverBus
+from v2.models import build_default_engine, build_omni_engine, build_wan21_card, build_wan_t2v_program
+from v2.request import DiffusionParams, OutputSpec, SamplingParams, TaskType, make_request
+from v2.runtime import AsyncEngine, DisaggregatedRunner, Engine, PoolSet, wan_t2v_disaggregated
+from v2.serving import OmniOpenAIServer
+from v2.transport import (
     InProcConnector,
     InProcKVConnector,
     ShmFakeConnector,
@@ -137,7 +137,7 @@ def _req(seed=1):
 
 
 def test_fleet_least_loaded_routes_around_busy_worker():
-    from mini_fastvideo.runtime import RequestState
+    from v2.runtime import RequestState
     A, B, cardA, cardB = _two_workers()
     fleet = LocalFleet("least_loaded")
     fleet.register("A", A, cardA); fleet.register("B", B, cardB)

v2/tests/test_serving_fixes.py

@@ -3,12 +3,12 @@ from __future__ import annotations
 
 import asyncio
 
-from mini_fastvideo.card import CostModel
-from mini_fastvideo.deploy import build_deployment_card
-from mini_fastvideo.models import build_default_engine, build_wan21_card, build_wan_t2v_program
-from mini_fastvideo.request import Cancelled, DiffusionParams, OutputSpec, TaskType, make_request
-from mini_fastvideo.runtime import AsyncEngine, PoolSet, wan_t2v_disaggregated
-from mini_fastvideo.transport import InProcConnector
+from v2.card import CostModel
+from v2.deploy import build_deployment_card
+from v2.models import build_default_engine, build_wan21_card, build_wan_t2v_program
+from v2.request import Cancelled, DiffusionParams, OutputSpec, TaskType, make_request
+from v2.runtime import AsyncEngine, PoolSet, wan_t2v_disaggregated
+from v2.transport import InProcConnector
 
 
 def _disagg_engine():
@@ -101,8 +101,8 @@ def test_offline_runner_honors_cancellation():
 
 
 def test_http_rejects_oversized_and_malformed_requests():
-    from mini_fastvideo.runtime import AsyncEngine as AE
-    from mini_fastvideo.serving import OmniOpenAIServer
+    from v2.runtime import AsyncEngine as AE
+    from v2.serving import OmniOpenAIServer
 
     async def status_of(host, port, raw: bytes) -> int:
         r, w = await asyncio.open_connection(host, port)

v2/tests/test_training.py

@@ -7,10 +7,10 @@ from __future__ import annotations
 
 import numpy as np
 
-from mini_fastvideo._enums import ConsistencyLevel
-from mini_fastvideo.models.wan21 import build_wan21_card
-from mini_fastvideo.models.wan_causal import build_wan_causal_card
-from mini_fastvideo.training import (
+from v2._enums import ConsistencyLevel
+from v2.models.wan21 import build_wan21_card
+from v2.models.wan_causal import build_wan_causal_card
+from v2.training import (
     WeightRole,
     build_diffusion_nft,
     build_dmd2,