mostlyai-engine 1.2.4__tar.gz → 1.3.1__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (50) hide show
  1. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/PKG-INFO +10 -20
  2. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/README.md +5 -5
  3. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/__init__.py +1 -1
  4. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/pyproject.toml +5 -38
  5. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/.gitignore +0 -0
  6. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/LICENSE +0 -0
  7. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_common.py +0 -0
  8. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_dtypes.py +0 -0
  9. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/__init__.py +0 -0
  10. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/language/__init__.py +0 -0
  11. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/language/categorical.py +0 -0
  12. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/language/datetime.py +0 -0
  13. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/language/numeric.py +0 -0
  14. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/language/text.py +0 -0
  15. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/tabular/__init__.py +0 -0
  16. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/tabular/categorical.py +0 -0
  17. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/tabular/character.py +0 -0
  18. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/tabular/datetime.py +0 -0
  19. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/tabular/itt.py +0 -0
  20. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/tabular/lat_long.py +0 -0
  21. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_encoding_types/tabular/numeric.py +0 -0
  22. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/__init__.py +0 -0
  23. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/common.py +0 -0
  24. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/encoding.py +0 -0
  25. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/engine/__init__.py +0 -0
  26. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/engine/base.py +0 -0
  27. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/engine/hf_engine.py +0 -0
  28. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/engine/vllm_engine.py +0 -0
  29. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/generation.py +0 -0
  30. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/lstm.py +0 -0
  31. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/tokenizer_utils.py +0 -0
  32. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/training.py +0 -0
  33. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_language/xgrammar_utils.py +0 -0
  34. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_memory.py +0 -0
  35. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_tabular/__init__.py +0 -0
  36. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_tabular/argn.py +0 -0
  37. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_tabular/common.py +0 -0
  38. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_tabular/encoding.py +0 -0
  39. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_tabular/fairness.py +0 -0
  40. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_tabular/generation.py +0 -0
  41. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_tabular/training.py +0 -0
  42. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_training_utils.py +0 -0
  43. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/_workspace.py +0 -0
  44. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/analysis.py +0 -0
  45. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/domain.py +0 -0
  46. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/encoding.py +0 -0
  47. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/generation.py +0 -0
  48. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/logging.py +0 -0
  49. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/splitting.py +0 -0
  50. {mostlyai_engine-1.2.4 → mostlyai_engine-1.3.1}/mostlyai/engine/training.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: mostlyai-engine
3
- Version: 1.2.4
3
+ Version: 1.3.1
4
4
  Summary: Synthetic Data Engine
5
5
  Project-URL: homepage, https://github.com/mostly-ai/mostlyai-engine
6
6
  Project-URL: repository, https://github.com/mostly-ai/mostlyai-engine
@@ -37,25 +37,15 @@ Requires-Dist: psutil<6,>=5.9.5
37
37
  Requires-Dist: pyarrow>=16.0.0
38
38
  Requires-Dist: setuptools<76,>=75.0.0
39
39
  Requires-Dist: tokenizers>=0.21.0
40
- Requires-Dist: torch==2.6.0; sys_platform == 'linux'
41
- Requires-Dist: torch>=2.6.0; sys_platform == 'darwin'
42
- Requires-Dist: torchaudio==2.6.0; sys_platform == 'linux'
43
- Requires-Dist: torchvision==0.21.0; sys_platform == 'linux'
40
+ Requires-Dist: torch<2.6.1,>=2.6.0
41
+ Requires-Dist: torchaudio<2.6.1,>=2.6.0
42
+ Requires-Dist: torchvision<0.21.1,>=0.21.0
44
43
  Requires-Dist: transformers>=4.51.0
45
44
  Requires-Dist: xgrammar>=0.1.16
46
- Provides-Extra: cpu
47
- Requires-Dist: torch==2.6.0+cpu; (sys_platform == 'linux') and extra == 'cpu'
48
- Requires-Dist: torch==2.6.0; (sys_platform != 'linux') and extra == 'cpu'
49
- Requires-Dist: torchaudio==2.6.0+cpu; (sys_platform == 'linux') and extra == 'cpu'
50
- Requires-Dist: torchvision==0.21.0+cpu; (sys_platform == 'linux') and extra == 'cpu'
51
45
  Provides-Extra: gpu
52
46
  Requires-Dist: bitsandbytes<0.43,>=0.42.0; (sys_platform == 'darwin') and extra == 'gpu'
53
47
  Requires-Dist: bitsandbytes<0.46,>=0.43.3; (sys_platform == 'linux') and extra == 'gpu'
54
- Requires-Dist: torch==2.6.0; extra == 'gpu'
55
- Requires-Dist: torchaudio==2.6.0; extra == 'gpu'
56
- Requires-Dist: torchvision==0.21.0; extra == 'gpu'
57
- Requires-Dist: vllm<=0.9,>=0.7.3; (sys_platform == 'darwin') and extra == 'gpu'
58
- Requires-Dist: vllm==0.8.3; (sys_platform == 'linux') and extra == 'gpu'
48
+ Requires-Dist: vllm==0.8.3; (sys_platform == 'linux' or sys_platform == 'darwin') and extra == 'gpu'
59
49
  Description-Content-Type: text/markdown
60
50
 
61
51
  # Synthetic Data Engine 💎
@@ -93,17 +83,17 @@ The latest release of `mostlyai-engine` can be installed via pip:
93
83
  pip install -U mostlyai-engine
94
84
  ```
95
85
 
96
- or alternatively for a CPU-only setup:
86
+ or alternatively for a GPU setup (needed for LLM finetuning and inference):
97
87
  ```bash
98
- pip install -U 'mostlyai-engine[cpu]' --extra-index-url https://download.pytorch.org/whl/cpu
88
+ pip install -U 'mostlyai-engine[gpu]'
99
89
  ```
100
90
 
101
- or alternatively for a GPU setup (needed for LLM finetuning):
91
+ On Linux, one can explicitly install the CPU-only variant of torch together with `mostlyai-engine`:
92
+
102
93
  ```bash
103
- pip install -U 'mostlyai-engine[gpu]'
94
+ pip install -U torch==2.6.0+cpu torchvision==0.21.0+cpu mostlyai-engine --extra-index-url https://download.pytorch.org/whl/cpu
104
95
  ```
105
96
 
106
-
107
97
  ## Quick start
108
98
 
109
99
  ### Tabular Model: flat data, without context
@@ -33,17 +33,17 @@ The latest release of `mostlyai-engine` can be installed via pip:
33
33
  pip install -U mostlyai-engine
34
34
  ```
35
35
 
36
- or alternatively for a CPU-only setup:
36
+ or alternatively for a GPU setup (needed for LLM finetuning and inference):
37
37
  ```bash
38
- pip install -U 'mostlyai-engine[cpu]' --extra-index-url https://download.pytorch.org/whl/cpu
38
+ pip install -U 'mostlyai-engine[gpu]'
39
39
  ```
40
40
 
41
- or alternatively for a GPU setup (needed for LLM finetuning):
41
+ On Linux, one can explicitly install the CPU-only variant of torch together with `mostlyai-engine`:
42
+
42
43
  ```bash
43
- pip install -U 'mostlyai-engine[gpu]'
44
+ pip install -U torch==2.6.0+cpu torchvision==0.21.0+cpu mostlyai-engine --extra-index-url https://download.pytorch.org/whl/cpu
44
45
  ```
45
46
 
46
-
47
47
  ## Quick start
48
48
 
49
49
  ### Tabular Model: flat data, without context
@@ -22,7 +22,7 @@ from mostlyai.engine.generation import generate
22
22
 
23
23
 
24
24
  __all__ = ["split", "analyze", "encode", "train", "generate", "init_logging"]
25
- __version__ = "1.2.4"
25
+ __version__ = "1.3.1"
26
26
 
27
27
  # suppress specific warning related to os.fork() in multi-threaded processes
28
28
  warnings.filterwarnings("ignore", category=DeprecationWarning, message=".*multi-threaded.*fork.*")
@@ -1,6 +1,6 @@
1
1
  [project]
2
2
  name = "mostlyai-engine"
3
- version = "1.2.4"
3
+ version = "1.3.1"
4
4
  description = "Synthetic Data Engine"
5
5
  authors = [{ name = "MOSTLY AI", email = "dev@mostly.ai" }]
6
6
  requires-python = ">=3.10"
@@ -40,27 +40,16 @@ dependencies = [
40
40
  "opacus>=1.5.2", # switch to 1.5.4 (once released) to allow numpy 2
41
41
  "xgrammar>=0.1.16", # for vllm 0.8.3 compatibility
42
42
  "json-repair>=0.30.0",
43
- "torch>=2.6.0; sys_platform == 'darwin'",
44
- "torch==2.6.0; sys_platform == 'linux'", # for vllm 0.8.3 compatibility
45
- "torchaudio==2.6.0; sys_platform == 'linux'", # for vllm 0.8.3 compatibility
46
- "torchvision==0.21.0; sys_platform == 'linux'" # for vllm 0.8.3 compatibility
43
+ "torch>=2.6.0,<2.6.1",
44
+ "torchaudio>=2.6.0,<2.6.1", # for vllm 0.8.3 compatibility
45
+ "torchvision>=0.21.0,<0.21.1" # for vllm 0.8.3 compatibility
47
46
  ]
48
47
 
49
48
  [project.optional-dependencies]
50
49
  gpu = [
51
- "torch==2.6.0", # for vllm 0.8.3 compatibility
52
- "torchaudio==2.6.0", # for vllm 0.8.3 compatibility
53
- "torchvision==0.21.0", # for vllm 0.8.3 compatibility
54
50
  "bitsandbytes>=0.42.0,<0.43; sys_platform == 'darwin'",
55
- "vllm>=0.7.3,<=0.9; sys_platform == 'darwin'",
56
51
  "bitsandbytes>=0.43.3,<0.46; sys_platform == 'linux'",
57
- "vllm==0.8.3; sys_platform == 'linux'",
58
- ]
59
- cpu = [
60
- "torch==2.6.0+cpu; sys_platform == 'linux'", # for vllm 0.8.3 compatibility
61
- "torchaudio==2.6.0+cpu; sys_platform == 'linux'", # for vllm 0.8.3 compatibility
62
- "torchvision==0.21.0+cpu; sys_platform == 'linux'", # for vllm 0.8.3 compatibility
63
- "torch==2.6.0; sys_platform != 'linux'",
52
+ "vllm==0.8.3; sys_platform == 'linux' or sys_platform == 'darwin'",
64
53
  ]
65
54
 
66
55
  [dependency-groups]
@@ -88,28 +77,6 @@ documentation = "https://mostly-ai.github.io/mostlyai-engine/"
88
77
 
89
78
  [tool.uv]
90
79
  default-groups = ["dev", "docs"]
91
- conflicts = [
92
- [
93
- { extra = "cpu" },
94
- { extra = "gpu" },
95
- ],
96
- ]
97
-
98
- [[tool.uv.index]]
99
- name = "pytorch-cpu"
100
- url = "https://download.pytorch.org/whl/cpu"
101
- explicit = true
102
-
103
- [tool.uv.sources]
104
- torch = [
105
- { index = "pytorch-cpu", extra = "cpu", marker = "sys_platform == 'linux'"},
106
- ]
107
- torchaudio = [
108
- { index = "pytorch-cpu", extra = "cpu", marker = "sys_platform == 'linux'"},
109
- ]
110
- torchvision = [
111
- { index = "pytorch-cpu", extra = "cpu", marker = "sys_platform == 'linux'"},
112
- ]
113
80
 
114
81
  [tool.hatch.build.targets.sdist]
115
82
  include = ["mostlyai/engine"]
File without changes