move test, update requirements

docs/zeta/ops/main.md

@@ -254,9 +254,7 @@ Returns:
 
 Let's explore some usage examples of the functions provided by the zeta library.
 
-#### 5.1 Example 1: Matrix Inverse Root using
-
- Eigen Method
+#### 5.1 Example 1: Matrix Inverse Root using Eigen Method
 
 In this example, we will compute the matrix inverse root of a symmetric positive definite matrix using the eigen method. We will use the following parameters:
 

requirements.txt

@@ -27,4 +27,5 @@ mkdocs
 mkdocs-material
 mkdocs-glightbox
 skypilot==0.4.1
-argparse
+argparse
+numexpr

tests/tokenizers/test_tokenmonster.py

@@ -49,15 +49,6 @@ def test_token_monster_new():
  assert tokenizer.vocab is not None
 
 
-def test_token_monster_save():
- tokenizer = TokenMonster("englishcode-32000-consistent-v1")
- tokenizer.save("/path/to/your/file") # replace with your actual file path
-
- # There's no direct way to assert the effect of this method as it doesn't return anything
- # and it doesn't change any accessible state of the TokenMonster object.
- # You might need to check manually if the file is saved correctly.
-
-
 def test_token_monster_export_yaml():
  tokenizer = TokenMonster("englishcode-32000-consistent-v1")
  yaml = tokenizer.export_yaml()

tests/utils/test_absmax.py

@@ -7,15 +7,15 @@
  quant, dequant = absmax_quantize(x)
  assert quant.dtype == torch.int8
  assert dequant.dtype == torch.float32
- assert torch.allclose(dequant, x, atol=1 / (2**7))
+ assert torch.allclose(dequant, x, atol=1e-5)
 
 
 def test_absmax_quantize_custom_bits():
  x = torch.randn(128)
  quant, dequant = absmax_quantize(x, bits=16)
  assert quant.dtype == torch.int8
  assert dequant.dtype == torch.float32
- assert torch.allclose(dequant, x, atol=1 / (2**15))
+ assert torch.allclose(dequant, x, atol=1e-5)
 
 
 def test_absmax_quantize_zero_tensor():
@@ -29,11 +29,11 @@
  x = torch.ones(128)
  quant, dequant = absmax_quantize(x)
  assert torch.all(quant == 2**7 - 1)
- assert torch.allclose(dequant, x, atol=1 / (2**7))
+ assert torch.allclose(dequant, x, atol=1e-5)
 
 
 def test_absmax_quantize_negative_tensor():
  x = -torch.ones(128)
  quant, dequant = absmax_quantize(x)
  assert torch.all(quant == -(2**7 - 1))
- assert torch.allclose(dequant, x, atol=1 / (2**7))
+ assert torch.allclose(dequant, x, atol=1e-5)

tests/utils/test_cast_tuple.py

@@ -31,12 +31,3 @@ def test_cast_tuple_parametrized(value, depth, expected):
 def test_cast_tuple_exception():
  with pytest.raises(TypeError):
  cast_tuple(5, "a")
-
-
-# Test with mock and monkeypatch
-def test_cast_tuple_with_mock_and_monkeypatch(monkeypatch):
- def mock_isinstance(val, t):
- return False
-
- monkeypatch.setattr("builtins.isinstance", mock_isinstance)
- assert cast_tuple((1, 2), 1) == ((1, 2),)

zeta/models/base.py

@@ -1,9 +1,8 @@
 from abc import ABC
 
-
 class BaseModel(ABC):
  def __init__(self, *args, **kwargs):
  pass
 
  def forward(self):
- pass
+ raise NotImplementedError

zeta/utils/main.py

@@ -458,7 +458,7 @@ def seek_all_images(img, channels=3):
 
 # tensor of shape (channels, frames, height, width) -> GIF
 def video_tensor_to_gift(tensor, path, duration=120, loop=0, optimize=True):
- images = map(T.ToPilImage(), tensor.unbind(dim=1))
+ images = map(T.ToPILImage(), tensor.unbind(dim=1))
  first_img, *rest_imgs = images
  first_img.save(
  path,