Datasets:
phanerozoic
commited on
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,67 @@
|
|
1 |
-
---
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- "en"
|
4 |
+
pretty_name: "MetaCoq Dataset"
|
5 |
+
tags:
|
6 |
+
- "theorem-proving"
|
7 |
+
- "formal-methods"
|
8 |
+
- "metacoq"
|
9 |
+
- "meta-programming"
|
10 |
+
- "coq"
|
11 |
+
license: "mit"
|
12 |
+
task_categories:
|
13 |
+
- "text-generation"
|
14 |
+
- "feature-extraction"
|
15 |
+
- "other"
|
16 |
+
---
|
17 |
+
|
18 |
+
# MetaCoq Dataset
|
19 |
+
|
20 |
+
## Dataset Description
|
21 |
+
The MetaCoq Dataset is derived from the MetaCoq repository, focusing on the formalization of Coq's meta-theory in the Coq proof assistant. This dataset processes .v files from the core theory directories to extract mathematical content in a structured format. This work builds upon the format established by Andreas Florath (@florath) in his Coq Facts, Propositions and Proofs dataset, providing a specialized view of the MetaCoq library with particular focus on meta-programming constructs.
|
22 |
+
|
23 |
+
### Dataset Structure
|
24 |
+
The dataset includes the following fields:
|
25 |
+
- fact: The mathematical statement body without type prefixes
|
26 |
+
- type: The statement type (Definition/Lemma/Theorem/etc., including MetaCoq-specific types like Quote Definition)
|
27 |
+
- library: The originating library (common/template-coq/pcuic/etc.)
|
28 |
+
- imports: The Require Import statements from the source file
|
29 |
+
- filename: The source file path within MetaCoq
|
30 |
+
- symbolic_name: The identifier of the mathematical object
|
31 |
+
- index_level: Sequential index for the dataset
|
32 |
+
|
33 |
+
### Example Row
|
34 |
+
fact: "forall {elt elt' : term}, red1 elt elt' -> red elt elt'"
|
35 |
+
type: "Lemma"
|
36 |
+
library: "pcuic"
|
37 |
+
imports: "From MetaCoq.PCUIC Require Import PCUICAst PCUICAstUtils PCUICInduction PCUICLiftSubst PCUICUnivSubst PCUICTyping PCUICReduction."
|
38 |
+
filename: "pcuic/theories/PCUICRedTypeIrrelevance.v"
|
39 |
+
symbolic_name: "red1_red"
|
40 |
+
index_level: 42
|
41 |
+
|
42 |
+
### Source Code
|
43 |
+
The dataset was generated using a custom Python script that processes core MetaCoq libraries. The extraction focuses on mathematical content while preserving the structure and relationships between definitions, lemmas, and their source files. Special handling is implemented for MetaCoq-specific constructs like Quote Definition and MetaCoq Run, as well as proper preservation of full proofs and statements.
|
44 |
+
|
45 |
+
### Coverage
|
46 |
+
The dataset includes content from the following MetaCoq directories:
|
47 |
+
- common/theories (Common utilities and definitions)
|
48 |
+
- template-coq/theories (Template-Coq implementation)
|
49 |
+
- pcuic/theories (Predicative Calculus of Inductive Constructions)
|
50 |
+
- safechecker/theories (Safe type-checker implementation)
|
51 |
+
- erasure/theories (Erasure translation)
|
52 |
+
|
53 |
+
### Usage
|
54 |
+
This dataset is designed for:
|
55 |
+
- Meta-theoretical Research: Analyzing formal proofs and definitions in Coq's meta-theory
|
56 |
+
- Machine Learning Applications: Training models for proof automation and tactic prediction
|
57 |
+
- Meta-programming Studies: Understanding patterns in quote/unquote operations and template programming
|
58 |
+
- Certified Programming: Studying certified meta-programming and proof checking
|
59 |
+
- Educational Purposes: Providing structured examples of meta-theoretical formalizations
|
60 |
+
|
61 |
+
### License
|
62 |
+
This dataset is distributed under the MIT license, aligning with the license of the original MetaCoq repository.
|
63 |
+
|
64 |
+
### Acknowledgments
|
65 |
+
- Original repository: MetaCoq (https://github.com/MetaCoq/metacoq)
|
66 |
+
- Inspiration: Hugging Face user Andreas Florath (@florath) and his comprehensive Coq dataset
|
67 |
+
- Related datasets: UniMath Dataset and Coq-HoTT Dataset
|