colonyml 0.1.1__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.
- colonyml-0.1.1/PKG-INFO +108 -0
- colonyml-0.1.1/README.md +93 -0
- colonyml-0.1.1/colonyml/__init__.py +1 -0
- colonyml-0.1.1/colonyml.egg-info/PKG-INFO +108 -0
- colonyml-0.1.1/colonyml.egg-info/SOURCES.txt +8 -0
- colonyml-0.1.1/colonyml.egg-info/dependency_links.txt +1 -0
- colonyml-0.1.1/colonyml.egg-info/top_level.txt +1 -0
- colonyml-0.1.1/pyproject.toml +3 -0
- colonyml-0.1.1/setup.cfg +4 -0
- colonyml-0.1.1/setup.py +16 -0
colonyml-0.1.1/PKG-INFO
ADDED
|
@@ -0,0 +1,108 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: colonyml
|
|
3
|
+
Version: 0.1.1
|
|
4
|
+
Summary: Zero-config distributed ML training for CPU clusters
|
|
5
|
+
Home-page: https://github.com/gjeevana27/colonyml
|
|
6
|
+
Author: Jeevana Gogineni
|
|
7
|
+
Requires-Python: >=3.10
|
|
8
|
+
Description-Content-Type: text/markdown
|
|
9
|
+
Dynamic: author
|
|
10
|
+
Dynamic: description
|
|
11
|
+
Dynamic: description-content-type
|
|
12
|
+
Dynamic: home-page
|
|
13
|
+
Dynamic: requires-python
|
|
14
|
+
Dynamic: summary
|
|
15
|
+
|
|
16
|
+
# ColonyML
|
|
17
|
+
|
|
18
|
+
> Zero-config distributed ML training for CPU clusters.
|
|
19
|
+
|
|
20
|
+
[](https://pypi.org/project/colonyml/)
|
|
21
|
+
[](https://www.python.org/downloads/)
|
|
22
|
+
[](https://opensource.org/licenses/MIT)
|
|
23
|
+
|
|
24
|
+
Pool any laptops into a training cluster — no IP addresses,
|
|
25
|
+
no config files, no GPU required. Just:
|
|
26
|
+
|
|
27
|
+
```bash
|
|
28
|
+
pip install colonyml
|
|
29
|
+
colonyml join
|
|
30
|
+
```
|
|
31
|
+
|
|
32
|
+
---
|
|
33
|
+
|
|
34
|
+
## The Problem
|
|
35
|
+
|
|
36
|
+
Training ML models on one laptop is slow. You might have
|
|
37
|
+
multiple machines sitting idle. Existing distributed training
|
|
38
|
+
tools require manual IP configuration, Linux, or expensive GPUs.
|
|
39
|
+
|
|
40
|
+
ColonyML fixes this. Any laptop. Any OS. Zero setup.
|
|
41
|
+
|
|
42
|
+
---
|
|
43
|
+
|
|
44
|
+
## How It Works
|
|
45
|
+
|
|
46
|
+
```
|
|
47
|
+
Laptop 1 → colonyml join (announces itself on network)
|
|
48
|
+
Laptop 2 → colonyml join (finds Laptop 1 automatically)
|
|
49
|
+
|
|
50
|
+
Both laptops train together:
|
|
51
|
+
- Auto-discovery via mDNS (like AirDrop for ML)
|
|
52
|
+
- Ring AllReduce for gradient synchronization
|
|
53
|
+
- Adaptive batch sizing based on CPU power
|
|
54
|
+
- Gradient compression over WiFi (up to 50x)
|
|
55
|
+
```
|
|
56
|
+
|
|
57
|
+
---
|
|
58
|
+
|
|
59
|
+
## Features
|
|
60
|
+
|
|
61
|
+
- Zero config — no IP addresses, no config files
|
|
62
|
+
- Auto-discovery — machines find each other via mDNS
|
|
63
|
+
- Ring AllReduce — efficient gradient averaging across nodes
|
|
64
|
+
- Adaptive scheduling — faster machines get bigger batches
|
|
65
|
+
- Gradient compression — up to 50x smaller over WiFi
|
|
66
|
+
- Any hardware — works on any CPU laptop, any OS
|
|
67
|
+
|
|
68
|
+
---
|
|
69
|
+
|
|
70
|
+
## Installation
|
|
71
|
+
|
|
72
|
+
```bash
|
|
73
|
+
pip install colonyml
|
|
74
|
+
```
|
|
75
|
+
|
|
76
|
+
---
|
|
77
|
+
|
|
78
|
+
## Status
|
|
79
|
+
|
|
80
|
+
Active Development — v0.1.0 is the initial release.
|
|
81
|
+
Core features being built now. Star the repo to follow progress.
|
|
82
|
+
|
|
83
|
+
---
|
|
84
|
+
|
|
85
|
+
## Roadmap
|
|
86
|
+
|
|
87
|
+
- v0.1.0 — PyPI release, project structure
|
|
88
|
+
- v0.2.0 — mDNS auto-discovery between machines
|
|
89
|
+
- v0.3.0 — gradient compression
|
|
90
|
+
- v0.4.0 — adaptive CPU-aware batch scheduler
|
|
91
|
+
- v0.5.0 — Ring AllReduce gradient synchronization
|
|
92
|
+
- v1.0.0 — stable release with full test coverage
|
|
93
|
+
|
|
94
|
+
---
|
|
95
|
+
|
|
96
|
+
## Author
|
|
97
|
+
|
|
98
|
+
Jeevana Sai Gogineni
|
|
99
|
+
MS Data Science · University of Maryland · GPA 4.0
|
|
100
|
+
|
|
101
|
+
GitHub: https://github.com/gjeevana27
|
|
102
|
+
LinkedIn: https://linkedin.com/in/jeevana-gogineni
|
|
103
|
+
|
|
104
|
+
---
|
|
105
|
+
|
|
106
|
+
## License
|
|
107
|
+
|
|
108
|
+
MIT License — free to use, modify, and distribute.
|
colonyml-0.1.1/README.md
ADDED
|
@@ -0,0 +1,93 @@
|
|
|
1
|
+
# ColonyML
|
|
2
|
+
|
|
3
|
+
> Zero-config distributed ML training for CPU clusters.
|
|
4
|
+
|
|
5
|
+
[](https://pypi.org/project/colonyml/)
|
|
6
|
+
[](https://www.python.org/downloads/)
|
|
7
|
+
[](https://opensource.org/licenses/MIT)
|
|
8
|
+
|
|
9
|
+
Pool any laptops into a training cluster — no IP addresses,
|
|
10
|
+
no config files, no GPU required. Just:
|
|
11
|
+
|
|
12
|
+
```bash
|
|
13
|
+
pip install colonyml
|
|
14
|
+
colonyml join
|
|
15
|
+
```
|
|
16
|
+
|
|
17
|
+
---
|
|
18
|
+
|
|
19
|
+
## The Problem
|
|
20
|
+
|
|
21
|
+
Training ML models on one laptop is slow. You might have
|
|
22
|
+
multiple machines sitting idle. Existing distributed training
|
|
23
|
+
tools require manual IP configuration, Linux, or expensive GPUs.
|
|
24
|
+
|
|
25
|
+
ColonyML fixes this. Any laptop. Any OS. Zero setup.
|
|
26
|
+
|
|
27
|
+
---
|
|
28
|
+
|
|
29
|
+
## How It Works
|
|
30
|
+
|
|
31
|
+
```
|
|
32
|
+
Laptop 1 → colonyml join (announces itself on network)
|
|
33
|
+
Laptop 2 → colonyml join (finds Laptop 1 automatically)
|
|
34
|
+
|
|
35
|
+
Both laptops train together:
|
|
36
|
+
- Auto-discovery via mDNS (like AirDrop for ML)
|
|
37
|
+
- Ring AllReduce for gradient synchronization
|
|
38
|
+
- Adaptive batch sizing based on CPU power
|
|
39
|
+
- Gradient compression over WiFi (up to 50x)
|
|
40
|
+
```
|
|
41
|
+
|
|
42
|
+
---
|
|
43
|
+
|
|
44
|
+
## Features
|
|
45
|
+
|
|
46
|
+
- Zero config — no IP addresses, no config files
|
|
47
|
+
- Auto-discovery — machines find each other via mDNS
|
|
48
|
+
- Ring AllReduce — efficient gradient averaging across nodes
|
|
49
|
+
- Adaptive scheduling — faster machines get bigger batches
|
|
50
|
+
- Gradient compression — up to 50x smaller over WiFi
|
|
51
|
+
- Any hardware — works on any CPU laptop, any OS
|
|
52
|
+
|
|
53
|
+
---
|
|
54
|
+
|
|
55
|
+
## Installation
|
|
56
|
+
|
|
57
|
+
```bash
|
|
58
|
+
pip install colonyml
|
|
59
|
+
```
|
|
60
|
+
|
|
61
|
+
---
|
|
62
|
+
|
|
63
|
+
## Status
|
|
64
|
+
|
|
65
|
+
Active Development — v0.1.0 is the initial release.
|
|
66
|
+
Core features being built now. Star the repo to follow progress.
|
|
67
|
+
|
|
68
|
+
---
|
|
69
|
+
|
|
70
|
+
## Roadmap
|
|
71
|
+
|
|
72
|
+
- v0.1.0 — PyPI release, project structure
|
|
73
|
+
- v0.2.0 — mDNS auto-discovery between machines
|
|
74
|
+
- v0.3.0 — gradient compression
|
|
75
|
+
- v0.4.0 — adaptive CPU-aware batch scheduler
|
|
76
|
+
- v0.5.0 — Ring AllReduce gradient synchronization
|
|
77
|
+
- v1.0.0 — stable release with full test coverage
|
|
78
|
+
|
|
79
|
+
---
|
|
80
|
+
|
|
81
|
+
## Author
|
|
82
|
+
|
|
83
|
+
Jeevana Sai Gogineni
|
|
84
|
+
MS Data Science · University of Maryland · GPA 4.0
|
|
85
|
+
|
|
86
|
+
GitHub: https://github.com/gjeevana27
|
|
87
|
+
LinkedIn: https://linkedin.com/in/jeevana-gogineni
|
|
88
|
+
|
|
89
|
+
---
|
|
90
|
+
|
|
91
|
+
## License
|
|
92
|
+
|
|
93
|
+
MIT License — free to use, modify, and distribute.
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
__version__ = "0.1.1"
|
|
@@ -0,0 +1,108 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: colonyml
|
|
3
|
+
Version: 0.1.1
|
|
4
|
+
Summary: Zero-config distributed ML training for CPU clusters
|
|
5
|
+
Home-page: https://github.com/gjeevana27/colonyml
|
|
6
|
+
Author: Jeevana Gogineni
|
|
7
|
+
Requires-Python: >=3.10
|
|
8
|
+
Description-Content-Type: text/markdown
|
|
9
|
+
Dynamic: author
|
|
10
|
+
Dynamic: description
|
|
11
|
+
Dynamic: description-content-type
|
|
12
|
+
Dynamic: home-page
|
|
13
|
+
Dynamic: requires-python
|
|
14
|
+
Dynamic: summary
|
|
15
|
+
|
|
16
|
+
# ColonyML
|
|
17
|
+
|
|
18
|
+
> Zero-config distributed ML training for CPU clusters.
|
|
19
|
+
|
|
20
|
+
[](https://pypi.org/project/colonyml/)
|
|
21
|
+
[](https://www.python.org/downloads/)
|
|
22
|
+
[](https://opensource.org/licenses/MIT)
|
|
23
|
+
|
|
24
|
+
Pool any laptops into a training cluster — no IP addresses,
|
|
25
|
+
no config files, no GPU required. Just:
|
|
26
|
+
|
|
27
|
+
```bash
|
|
28
|
+
pip install colonyml
|
|
29
|
+
colonyml join
|
|
30
|
+
```
|
|
31
|
+
|
|
32
|
+
---
|
|
33
|
+
|
|
34
|
+
## The Problem
|
|
35
|
+
|
|
36
|
+
Training ML models on one laptop is slow. You might have
|
|
37
|
+
multiple machines sitting idle. Existing distributed training
|
|
38
|
+
tools require manual IP configuration, Linux, or expensive GPUs.
|
|
39
|
+
|
|
40
|
+
ColonyML fixes this. Any laptop. Any OS. Zero setup.
|
|
41
|
+
|
|
42
|
+
---
|
|
43
|
+
|
|
44
|
+
## How It Works
|
|
45
|
+
|
|
46
|
+
```
|
|
47
|
+
Laptop 1 → colonyml join (announces itself on network)
|
|
48
|
+
Laptop 2 → colonyml join (finds Laptop 1 automatically)
|
|
49
|
+
|
|
50
|
+
Both laptops train together:
|
|
51
|
+
- Auto-discovery via mDNS (like AirDrop for ML)
|
|
52
|
+
- Ring AllReduce for gradient synchronization
|
|
53
|
+
- Adaptive batch sizing based on CPU power
|
|
54
|
+
- Gradient compression over WiFi (up to 50x)
|
|
55
|
+
```
|
|
56
|
+
|
|
57
|
+
---
|
|
58
|
+
|
|
59
|
+
## Features
|
|
60
|
+
|
|
61
|
+
- Zero config — no IP addresses, no config files
|
|
62
|
+
- Auto-discovery — machines find each other via mDNS
|
|
63
|
+
- Ring AllReduce — efficient gradient averaging across nodes
|
|
64
|
+
- Adaptive scheduling — faster machines get bigger batches
|
|
65
|
+
- Gradient compression — up to 50x smaller over WiFi
|
|
66
|
+
- Any hardware — works on any CPU laptop, any OS
|
|
67
|
+
|
|
68
|
+
---
|
|
69
|
+
|
|
70
|
+
## Installation
|
|
71
|
+
|
|
72
|
+
```bash
|
|
73
|
+
pip install colonyml
|
|
74
|
+
```
|
|
75
|
+
|
|
76
|
+
---
|
|
77
|
+
|
|
78
|
+
## Status
|
|
79
|
+
|
|
80
|
+
Active Development — v0.1.0 is the initial release.
|
|
81
|
+
Core features being built now. Star the repo to follow progress.
|
|
82
|
+
|
|
83
|
+
---
|
|
84
|
+
|
|
85
|
+
## Roadmap
|
|
86
|
+
|
|
87
|
+
- v0.1.0 — PyPI release, project structure
|
|
88
|
+
- v0.2.0 — mDNS auto-discovery between machines
|
|
89
|
+
- v0.3.0 — gradient compression
|
|
90
|
+
- v0.4.0 — adaptive CPU-aware batch scheduler
|
|
91
|
+
- v0.5.0 — Ring AllReduce gradient synchronization
|
|
92
|
+
- v1.0.0 — stable release with full test coverage
|
|
93
|
+
|
|
94
|
+
---
|
|
95
|
+
|
|
96
|
+
## Author
|
|
97
|
+
|
|
98
|
+
Jeevana Sai Gogineni
|
|
99
|
+
MS Data Science · University of Maryland · GPA 4.0
|
|
100
|
+
|
|
101
|
+
GitHub: https://github.com/gjeevana27
|
|
102
|
+
LinkedIn: https://linkedin.com/in/jeevana-gogineni
|
|
103
|
+
|
|
104
|
+
---
|
|
105
|
+
|
|
106
|
+
## License
|
|
107
|
+
|
|
108
|
+
MIT License — free to use, modify, and distribute.
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
colonyml
|
colonyml-0.1.1/setup.cfg
ADDED
colonyml-0.1.1/setup.py
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
from setuptools import setup, find_packages
|
|
2
|
+
|
|
3
|
+
with open("README.md", "r", encoding="utf-8") as f:
|
|
4
|
+
long_description = f.read()
|
|
5
|
+
|
|
6
|
+
setup(
|
|
7
|
+
name="colonyml",
|
|
8
|
+
version="0.1.1",
|
|
9
|
+
author="Jeevana Gogineni",
|
|
10
|
+
description="Zero-config distributed ML training for CPU clusters",
|
|
11
|
+
long_description=long_description,
|
|
12
|
+
long_description_content_type="text/markdown",
|
|
13
|
+
url="https://github.com/gjeevana27/colonyml",
|
|
14
|
+
packages=find_packages(),
|
|
15
|
+
python_requires=">=3.10",
|
|
16
|
+
)
|