Update paths to data files (#1996)

* Update paths to data files

* address numerical issues due to new jax release

* fix the size of the new SP500 file

Author: fehiepsi

numpyro/examples/datasets.py

@@ -28,7 +28,8 @@
 dset = namedtuple("dset", ["name", "urls"])
 
 BASEBALL = dset(
-    "baseball", ["https://d2hg8soec8ck9v.cloudfront.net/datasets/EfronMorrisBB.txt"]
+    "baseball",
+    ["https://github.com/pyro-ppl/datasets/blob/master/EfronMorrisBB.txt?raw=true"],
 )
 
 BOSTON_HOUSING = dset(
@@ -37,7 +38,7 @@
 )
 
 COVTYPE = dset(
-    "covtype", ["https://d2hg8soec8ck9v.cloudfront.net/datasets/covtype.zip"]
+    "covtype", ["https://github.com/pyro-ppl/datasets/blob/master/covtype.npz?raw=true"]
 )
 
 DIPPER_VOLE = dset(
@@ -48,26 +49,32 @@
 MNIST = dset(
     "mnist",
     [
-        "https://d2hg8soec8ck9v.cloudfront.net/datasets/mnist/train-images-idx3-ubyte.gz",
-        "https://d2hg8soec8ck9v.cloudfront.net/datasets/mnist/train-labels-idx1-ubyte.gz",
-        "https://d2hg8soec8ck9v.cloudfront.net/datasets/mnist/t10k-images-idx3-ubyte.gz",
-        "https://d2hg8soec8ck9v.cloudfront.net/datasets/mnist/t10k-labels-idx1-ubyte.gz",
+        "https://github.com/pyro-ppl/datasets/blob/master/mnist/train-images-idx3-ubyte.gz?raw=true",
+        "https://github.com/pyro-ppl/datasets/blob/master/mnist/train-labels-idx1-ubyte.gz?raw=true",
+        "https://github.com/pyro-ppl/datasets/blob/master/mnist/t10k-images-idx3-ubyte.gz?raw=true",
+        "https://github.com/pyro-ppl/datasets/blob/master/mnist/t10k-labels-idx1-ubyte.gz?raw=true",
     ],
 )
 
-SP500 = dset("SP500", ["https://d2hg8soec8ck9v.cloudfront.net/datasets/SP500.csv"])
+SP500 = dset(
+    "SP500", ["https://github.com/pyro-ppl/datasets/blob/master/SP500.csv?raw=true"]
+)
 
 UCBADMIT = dset(
-    "ucbadmit", ["https://d2hg8soec8ck9v.cloudfront.net/datasets/UCBadmit.csv"]
+    "ucbadmit",
+    ["https://github.com/pyro-ppl/datasets/blob/master/UCBadmit.csv?raw=true"],
 )
 
 LYNXHARE = dset(
-    "lynxhare", ["https://d2hg8soec8ck9v.cloudfront.net/datasets/LynxHare.txt"]
+    "lynxhare",
+    ["https://github.com/pyro-ppl/datasets/blob/master/LynxHare.txt?raw=true"],
 )
 
 JSB_CHORALES = dset(
     "jsb_chorales",
-    ["https://d2hg8soec8ck9v.cloudfront.net/datasets/polyphonic/jsb_chorales.pickle"],
+    [
+        "https://github.com/pyro-ppl/datasets/blob/master/polyphonic/jsb_chorales.pickle?raw=true"
+    ],
 )
 
 HIGGS = dset(
@@ -129,7 +136,7 @@ def _load_boston_housing():
 def _load_covtype():
     _download(COVTYPE)
 
-    file_path = os.path.join(DATA_DIR, "covtype.zip")
+    file_path = os.path.join(DATA_DIR, "covtype.npz")
     data = np.load(file_path)
 
     return {"train": (data["data"], data["target"])}

test/infer/test_autoguide.py

@@ -405,7 +405,7 @@ def model(x, y):
         b = numpyro.sample("b", dist.Normal(0, 10).expand([3]).to_event())
         mu = a + b[0] * x + b[1] * x**2 + b[2] * x**3
         with numpyro.plate("N", len(x)):
-            numpyro.sample("y", dist.Normal(mu, 0.001), obs=y)
+            numpyro.sample("y", dist.Normal(mu, 0.00001), obs=y)
 
     x = random.normal(random.PRNGKey(0), (3,))
     y = 1 + 2 * x + 3 * x**2 + 4 * x**3

test/test_distributions.py

@@ -1586,7 +1586,7 @@ def test_entropy_categorical():
     probs = _to_probs_multinom(logits)
     sp_dist = osp.multinomial(1, probs)
     for jax_dist in [dist.CategoricalLogits(logits), dist.CategoricalProbs(probs)]:
-        assert_allclose(jax_dist.entropy(), sp_dist.entropy())
+        assert_allclose(jax_dist.entropy(), sp_dist.entropy(), rtol=1e-6, atol=1e-6)
 
 
 def test_mixture_log_prob():

test/test_example_utils.py

@@ -42,7 +42,7 @@ def mean_pixels(i, mean_pix):
 def test_sp500_data_load():
     _, fetch = load_dataset(SP500, split="train", shuffle=False)
     date, value = fetch()
-    assert jnp.shape(date) == jnp.shape(date) == (2427,)
+    assert jnp.shape(date) == jnp.shape(date) == (2517,)
 
 
 def test_jsb_chorales():

test/test_transforms.py

@@ -322,7 +322,7 @@ def test_bijective_transforms(transform, shape):
     if isinstance(transform, less_stable_transforms):
         atol = 1e-2
     elif isinstance(transform, (L1BallTransform, RecursiveLinearTransform)):
-        atol = 0.1
+        atol = 0.2
     assert jnp.allclose(x1, x2, atol=atol)
 
     log_abs_det_jacobian = transform.log_abs_det_jacobian(x1, y)