tirex-mirror 2025.9.10__tar.gz → 2025.10.1__tar.gz

{tirex_mirror-2025.9.10/src/tirex_mirror.egg-info → tirex_mirror-2025.10.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tirex-mirror
-Version: 2025.9.10
+Version: 2025.10.1
 Summary: Unofficial mirror of NX-AI/tirex for packaging
 Author-email: Arpad Rozsas <rozsasarpi@gmail.com>
 License: NXAI COMMUNITY LICENSE AGREEMENT
@@ -80,7 +80,7 @@ Requires-Dist: gluonts; extra == "gluonts"
 Provides-Extra: hfdataset
 Requires-Dist: datasets; extra == "hfdataset"
 Provides-Extra: test
-Requires-Dist: fev; extra == "test"
+Requires-Dist: fev>=0.6.0; extra == "test"
 Requires-Dist: pytest; extra == "test"
 Provides-Extra: all
 Requires-Dist: xlstm; extra == "all"
@@ -90,7 +90,7 @@ Requires-Dist: matplotlib; extra == "all"
 Requires-Dist: gluonts; extra == "all"
 Requires-Dist: datasets; extra == "all"
 Requires-Dist: pytest; extra == "all"
-Requires-Dist: fev; extra == "all"
+Requires-Dist: fev>=0.6.0; extra == "all"
 Dynamic: license-file
 
 # tirex-mirror

{tirex_mirror-2025.9.10 → tirex_mirror-2025.10.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "tirex-mirror"
-version = "2025.09.10"
+version = "2025.10.01"
 description = "Unofficial mirror of NX-AI/tirex for packaging"
 readme = "README.md"
 requires-python = ">=3.11"
@@ -27,8 +27,8 @@ cuda = [ "xlstm", "ninja",]
 notebooks = [ "ipykernel", "matplotlib",]
 gluonts = [ "gluonts",]
 hfdataset = [ "datasets",]
-test = [ "fev", "pytest",]
-all = [ "xlstm", "ninja", "ipykernel", "matplotlib", "gluonts", "datasets", "pytest", "fev",]
+test = [ "fev>=0.6.0", "pytest",]
+all = [ "xlstm", "ninja", "ipykernel", "matplotlib", "gluonts", "datasets", "pytest", "fev>=0.6.0",]
 
 [tool.docformatter]
 diff = false

{tirex_mirror-2025.9.10 → tirex_mirror-2025.10.1/src/tirex_mirror.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tirex-mirror
-Version: 2025.9.10
+Version: 2025.10.1
 Summary: Unofficial mirror of NX-AI/tirex for packaging
 Author-email: Arpad Rozsas <rozsasarpi@gmail.com>
 License: NXAI COMMUNITY LICENSE AGREEMENT
@@ -80,7 +80,7 @@ Requires-Dist: gluonts; extra == "gluonts"
 Provides-Extra: hfdataset
 Requires-Dist: datasets; extra == "hfdataset"
 Provides-Extra: test
-Requires-Dist: fev; extra == "test"
+Requires-Dist: fev>=0.6.0; extra == "test"
 Requires-Dist: pytest; extra == "test"
 Provides-Extra: all
 Requires-Dist: xlstm; extra == "all"
@@ -90,7 +90,7 @@ Requires-Dist: matplotlib; extra == "all"
 Requires-Dist: gluonts; extra == "all"
 Requires-Dist: datasets; extra == "all"
 Requires-Dist: pytest; extra == "all"
-Requires-Dist: fev; extra == "all"
+Requires-Dist: fev>=0.6.0; extra == "all"
 Dynamic: license-file
 
 # tirex-mirror

{tirex_mirror-2025.9.10 → tirex_mirror-2025.10.1}/src/tirex_mirror.egg-info/requires.txt

@@ -13,7 +13,7 @@ matplotlib
 gluonts
 datasets
 pytest
-fev
+fev>=0.6.0
 
 [cuda]
 xlstm
@@ -30,5 +30,5 @@ ipykernel
 matplotlib
 
 [test]
-fev
+fev>=0.6.0
 pytest

{tirex_mirror-2025.9.10 → tirex_mirror-2025.10.1}/tests/test_chronos_zs.py

@@ -14,22 +14,30 @@ def geometric_mean(s):
 
 
 def eval_task(model, task):
-    past_data, _ = task.get_input_data(trust_remote_code=True)
-    quantile_levels = task.quantile_levels
-    past_data = past_data.with_format("torch").cast_column(
-        task.target_column, datasets.Sequence(datasets.Value("float32"))
-    )[task.target_column]
-    loaded_data = [t for t in past_data]
-
-    start_time = time.monotonic()
-    quantiles, means = model.forecast(loaded_data, quantile_levels=quantile_levels, prediction_length=task.horizon)
-    inference_time = time.monotonic() - start_time
-    predictions_dict = {"predictions": means}
-    for idx, level in enumerate(quantile_levels):
-        predictions_dict[str(level)] = quantiles[:, :, idx]  # [num_items, horizon]
-
-    predictions = datasets.Dataset.from_dict(predictions_dict)
-    return predictions, inference_time
+    inference_time = 0.0
+    predictions_per_window = []
+    for window in task.iter_windows(trust_remote_code=True):
+        past_data, _ = fev.convert_input_data(window, adapter="datasets", as_univariate=True)
+        past_data = past_data.with_format("torch").cast_column("target", datasets.Sequence(datasets.Value("float32")))
+        loaded_targets = [t for t in past_data["target"]]
+
+        start_time = time.monotonic()
+        quantiles, means = model.forecast(
+            loaded_targets, quantile_levels=task.quantile_levels, prediction_length=task.horizon
+        )
+        inference_time += time.monotonic() - start_time
+
+        predictions_dict = {"predictions": means}
+        for idx, level in enumerate(task.quantile_levels):
+            predictions_dict[str(level)] = quantiles[:, :, idx]
+
+        predictions_per_window.append(
+            fev.combine_univariate_predictions_to_multivariate(
+                datasets.Dataset.from_dict(predictions_dict), target_columns=task.target_columns
+            )
+        )
+
+    return predictions_per_window, inference_time
 
 
 @pytest.fixture

tirex_mirror-2025.10.1/tests/test_forecast.py

@@ -0,0 +1,48 @@
+from pathlib import Path
+
+import numpy as np
+import pytest
+import torch
+
+from tirex import ForecastModel, load_model
+
+
+def load_tensor_from_txt_file(path):
+    base_path = Path(__file__).parent.resolve() / "data"
+    return torch.from_numpy(np.genfromtxt(base_path / path, dtype=np.float32))
+
+
+def load_tensor_from_pt_file(path):
+    base_path = Path(__file__).parent.resolve() / "data"
+    return torch.load(base_path / path)
+
+
+@pytest.fixture
+def tirex_model() -> ForecastModel:
+    return load_model("NX-AI/TiRex")
+
+
+def test_forecast_air_traffic(tirex_model):
+    context = load_tensor_from_txt_file("air_passengers.csv")[:-12]
+
+    quantiles, mean = tirex_model.forecast(context, prediction_length=24)
+
+    ref_mean = load_tensor_from_txt_file("air_passengers_forecast_ref.csv").unsqueeze(0)
+    ref_quantiles = load_tensor_from_pt_file("air_passengers_quantiles_ref.pt")
+
+    # default rtol & atol for bfloat16
+    torch.testing.assert_close(mean, ref_mean, rtol=1.6e-2, atol=1e-5)
+    torch.testing.assert_close(quantiles, ref_quantiles, rtol=1.6e-2, atol=1e-5)
+
+
+def test_forecast_seattle_5T(tirex_model):
+    context = load_tensor_from_txt_file("loop_seattle_5T.csv")[:-512]
+
+    quantiles, mean = tirex_model.forecast(context, prediction_length=768)
+
+    ref_mean = load_tensor_from_txt_file("loop_seattle_5T_forecast_ref.csv").unsqueeze(0)
+    ref_quantiles = load_tensor_from_pt_file("loop_seattle_5T_quantiles_ref.pt")
+
+    # default rtol & atol for bfloat16
+    torch.testing.assert_close(mean, ref_mean, rtol=1.6e-2, atol=1e-5)
+    torch.testing.assert_close(quantiles, ref_quantiles, rtol=1.6e-2, atol=1e-5)

tirex_mirror-2025.9.10/tests/test_forecast.py

@@ -1,37 +0,0 @@
-from pathlib import Path
-
-import numpy as np
-import pytest
-import torch
-
-from tirex import ForecastModel, load_model
-
-
-def load_tensor_from_file(path):
-    base_path = Path(__file__).parent.resolve() / "data"
-    return torch.from_numpy(np.genfromtxt(base_path / path, dtype=np.float32))
-
-
-@pytest.fixture
-def tirex_model() -> ForecastModel:
-    return load_model("NX-AI/TiRex")
-
-
-def test_forecast_air_traffic(tirex_model):
-    context = load_tensor_from_file("air_passengers.csv")[:-12]
-
-    quantiles, mean = tirex_model.forecast(context, prediction_length=24)
-
-    ref_data = load_tensor_from_file("air_passengers_forecast_ref.csv").unsqueeze(0)
-
-    torch.testing.assert_close(mean, ref_data, rtol=1.6e-2, atol=1e-5)  # default rtol & atol for bfloat16
-
-
-def test_forecast_seattle_5T(tirex_model):
-    context = load_tensor_from_file("loop_seattle_5T.csv")[:-512]
-
-    quantiles, mean = tirex_model.forecast(context, prediction_length=768)
-
-    ref_data = load_tensor_from_file("loop_seattle_5T_forecast_ref.csv").unsqueeze(0)
-
-    torch.testing.assert_close(mean, ref_data, rtol=1.6e-2, atol=1e-5)  # default rtol & atol for bfloat16