embedprobe 0.0.0__py3-none-any.whl
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.
embedprobe/__init__.py
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: embedprobe
|
|
3
|
+
Version: 0.0.0
|
|
4
|
+
Summary: A diagnostic toolkit for evaluating and selecting language-model embedding spaces.
|
|
5
|
+
Project-URL: Homepage, https://github.com/Sainath26/embedprobe
|
|
6
|
+
Author-email: Harish Sainath S <sainathharish036@gmail.com>
|
|
7
|
+
License: MIT
|
|
8
|
+
Keywords: diagnostics,embeddings,evaluation,model-selection,nlp,transformers
|
|
9
|
+
Classifier: Development Status :: 1 - Planning
|
|
10
|
+
Classifier: Intended Audience :: Science/Research
|
|
11
|
+
Classifier: License :: OSI Approved :: MIT License
|
|
12
|
+
Classifier: Programming Language :: Python :: 3
|
|
13
|
+
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
|
|
14
|
+
Requires-Python: >=3.9
|
|
15
|
+
Description-Content-Type: text/markdown
|
|
16
|
+
|
|
17
|
+
# embedprobe
|
|
18
|
+
|
|
19
|
+
A diagnostic toolkit for evaluating and selecting language-model embedding spaces.
|
|
20
|
+
|
|
21
|
+
> **Status: early development.** This is a placeholder release to reserve the name.
|
|
22
|
+
> The first working version is coming soon.
|
|
23
|
+
|
|
24
|
+
## What it will do
|
|
25
|
+
|
|
26
|
+
`embedprobe` helps you choose the right embedding model for _your_ downstream task
|
|
27
|
+
by diagnosing _why_ models succeed or fail on _your_ data and language pairs, going
|
|
28
|
+
beyond aggregate leaderboard scores. It probes an embedding space across four levels:
|
|
29
|
+
|
|
30
|
+
- **Signal-to-noise separability** — how cleanly true pairs separate from noise.
|
|
31
|
+
- **Retrieval performance** — Recall@K, MRR, and cumulative-match analysis.
|
|
32
|
+
- **Topic-level structure** — UMAP projections and topic-cohesion heatmaps.
|
|
33
|
+
- **Error categorisation** — a taxonomy of retrieval misses (lexical / semantic / topic-boundary).
|
|
34
|
+
|
|
35
|
+
## Install
|
|
36
|
+
|
|
37
|
+
```bash
|
|
38
|
+
pip install embedprobe
|
|
39
|
+
```
|
|
40
|
+
|
|
41
|
+
## License
|
|
42
|
+
|
|
43
|
+
MIT
|
|
@@ -0,0 +1,4 @@
|
|
|
1
|
+
embedprobe/__init__.py,sha256=zU49oKDXQJSNtJGC6WeOet49jzSVTky-Scw6oykFlCE,114
|
|
2
|
+
embedprobe-0.0.0.dist-info/METADATA,sha256=nJPR9au2_kOrTEH65Zz4lhPuzmUy8Ape--zkQ7Ppwnw,1596
|
|
3
|
+
embedprobe-0.0.0.dist-info/WHEEL,sha256=mffPy8wBnZQn2VnJUU5jE99KsxaSfiyMHV9Yt0aLVxs,87
|
|
4
|
+
embedprobe-0.0.0.dist-info/RECORD,,
|