Datasets:
Commit
•
e8b9370
1
Parent(s):
d5c4660
Update README.md (#1)
Browse files- Update README.md (32c07bfd941d82f8c5b4d20952845d65011fffe6)
Co-authored-by: Gabriel Ebner <gebnermsft@users.noreply.huggingface.co>
README.md
CHANGED
@@ -19,11 +19,11 @@ language:
|
|
19 |
# Proof Oriented Programming with AI (PoPAI) - FStarDataSet
|
20 |
|
21 |
This dataset contains programs and proofs in [F* proof-oriented programming language](https://fstar-lang.org/).
|
22 |
-
The data, proposed in [Towards Neural Synthesis for SMT-Assisted Proof-Oriented Programming](https://arxiv.org/pdf/2405.01787)
|
23 |
is an archive of source code, build artifacts, and metadata assembled from eight different F⋆-based open source projects on GitHub.
|
24 |
|
25 |
## Primary-Objective
|
26 |
-
This dataset's primary objective is to train and evaluate Proof-oriented Programming with AI (PoPAI, in short). Given a specification of a program and proof
|
27 |
the objective of a AI model is to synthesize the implemantation (see [below](#usage) for details about the usage of this dataset, including the input and output).
|
28 |
|
29 |
## Data Format
|
@@ -35,27 +35,27 @@ Each of the examples in this dataset are organized as dictionaries with the foll
|
|
35 |
"original_source_type": <str: actual source type, to be used for type checking>,
|
36 |
"source_type": <str: modified source type, to be used to formulate prompt>,
|
37 |
"source_definition": <str: target definition>,
|
38 |
-
"source": <dict: contains metadata
|
39 |
"source_range": <dict: metadata containing start and end lines and columns of this definition in the source file>,
|
40 |
"file_context": <str: extracted file context upto the point of current definition>,
|
41 |
"dependencies": <dict: build dependencies for this file>,
|
42 |
-
"opens_and_abbrevs": <list[dict]: List of opened modules and abbreviated modules in the file,
|
43 |
-
"vconfig": <dict:
|
44 |
-
"interleaved": <bool: whether this definition is interleaved
|
45 |
"verbose_type": <str: the verbose type of this definition as resolved by the type checker>,
|
46 |
"effect": <str: effect>,
|
47 |
"effect_flags": <list[str]: any effect flags>,
|
48 |
-
"mutual_with": <list: if this definition is mutually recursive with
|
49 |
"ideal_premises": <list[str]: Other definitions that are used in the ground truth definition>,
|
50 |
"proof_features": <list[str]>,
|
51 |
"is_simple_lemma": <bool/null>,
|
52 |
-
"is_div": <bool: if this definition
|
53 |
"is_proof": <bool>,
|
54 |
"is_simply_typed": <bool>,
|
55 |
"is_type": <bool/null>,
|
56 |
"partial_definition": <str>,
|
57 |
"completed_definiton": <str>,
|
58 |
-
"isa_cross_project_example": <bool: if this example belongs to cross-project evaluation set>
|
59 |
}
|
60 |
```
|
61 |
|
@@ -90,14 +90,14 @@ Download the source code and the `helpers.zip` file from the release.
|
|
90 |
|
91 |
## Troubleshooting
|
92 |
The attached binaries in the evaluator (i.e., `fstar.exe` and `z3`) are built on
|
93 |
-
**`Ubuntu 20.04.6 LTS (GNU/Linux 5.4.0-189-generic x86_64)
|
94 |
If any of the binaries do not work properly, build F* from [commit: f3b4db2ebce90020acbbbe1b4ea0d05d3e69ad6c](https://github.com/FStarLang/FStar/commit/f3b4db2ebce90020acbbbe1b4ea0d05d3e69ad6c)
|
95 |
-
from the [F* repository](https://github.com/FStarLang/FStar), using
|
96 |
|
97 |
|
98 |
# Data Source
|
99 |
-
The raw data in this project are collected from eight open-source F*
|
100 |
-
1. [FStar](https://github.com/FStarLang/FStar): The F⋆ compiler itself, including its standard
|
101 |
2. [Karamel](https://github.com/FStarLang/karamel): A transpiler from a subset of F⋆ called Low* to C, including libraries to work with a model of C types and control structures, e.g., for- and while-loops.
|
102 |
3. [EverParse](https://github.com/project-everest/everparse): A parser generator for binary formats, used in various large scale systems, e.g., the Windows kernel.
|
103 |
4. [HACL*](https://github.com/hacl-star/hacl-star): A library of verified cryptographic algorithms, including ValeCrypt, a library of verified assembly code, as well as EverCrypt, a cryptographic provider, including code deployed in Linux, Firefox, and Python.
|
|
|
19 |
# Proof Oriented Programming with AI (PoPAI) - FStarDataSet
|
20 |
|
21 |
This dataset contains programs and proofs in [F* proof-oriented programming language](https://fstar-lang.org/).
|
22 |
+
The data, proposed in [Towards Neural Synthesis for SMT-Assisted Proof-Oriented Programming](https://arxiv.org/pdf/2405.01787),
|
23 |
is an archive of source code, build artifacts, and metadata assembled from eight different F⋆-based open source projects on GitHub.
|
24 |
|
25 |
## Primary-Objective
|
26 |
+
This dataset's primary objective is to train and evaluate Proof-oriented Programming with AI (PoPAI, in short). Given a specification of a program and proof in F*,
|
27 |
the objective of a AI model is to synthesize the implemantation (see [below](#usage) for details about the usage of this dataset, including the input and output).
|
28 |
|
29 |
## Data Format
|
|
|
35 |
"original_source_type": <str: actual source type, to be used for type checking>,
|
36 |
"source_type": <str: modified source type, to be used to formulate prompt>,
|
37 |
"source_definition": <str: target definition>,
|
38 |
+
"source": <dict: contains metadata about the source of this example, including project_name, git url, git sha, etc.>,
|
39 |
"source_range": <dict: metadata containing start and end lines and columns of this definition in the source file>,
|
40 |
"file_context": <str: extracted file context upto the point of current definition>,
|
41 |
"dependencies": <dict: build dependencies for this file>,
|
42 |
+
"opens_and_abbrevs": <list[dict]: List of opened modules and abbreviated modules in the file, necessary for evaluation.>,
|
43 |
+
"vconfig": <dict: SMT solver flags for this definition>,
|
44 |
+
"interleaved": <bool: whether this definition is interleaved from the interface file>,
|
45 |
"verbose_type": <str: the verbose type of this definition as resolved by the type checker>,
|
46 |
"effect": <str: effect>,
|
47 |
"effect_flags": <list[str]: any effect flags>,
|
48 |
+
"mutual_with": <list: if this definition is mutually recursive with another, list of those names>,
|
49 |
"ideal_premises": <list[str]: Other definitions that are used in the ground truth definition>,
|
50 |
"proof_features": <list[str]>,
|
51 |
"is_simple_lemma": <bool/null>,
|
52 |
+
"is_div": <bool: if this definition has the divergent effect>,
|
53 |
"is_proof": <bool>,
|
54 |
"is_simply_typed": <bool>,
|
55 |
"is_type": <bool/null>,
|
56 |
"partial_definition": <str>,
|
57 |
"completed_definiton": <str>,
|
58 |
+
"isa_cross_project_example": <bool: if this example belongs to the cross-project evaluation set>
|
59 |
}
|
60 |
```
|
61 |
|
|
|
90 |
|
91 |
## Troubleshooting
|
92 |
The attached binaries in the evaluator (i.e., `fstar.exe` and `z3`) are built on
|
93 |
+
**`Ubuntu 20.04.6 LTS (GNU/Linux 5.4.0-189-generic x86_64)`**, **`gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2)`**, **`OCaml 4.12.0`**.
|
94 |
If any of the binaries do not work properly, build F* from [commit: f3b4db2ebce90020acbbbe1b4ea0d05d3e69ad6c](https://github.com/FStarLang/FStar/commit/f3b4db2ebce90020acbbbe1b4ea0d05d3e69ad6c)
|
95 |
+
from the [F* repository](https://github.com/FStarLang/FStar), using the [installation guide](https://github.com/FStarLang/FStar/blob/master/INSTALL.md).
|
96 |
|
97 |
|
98 |
# Data Source
|
99 |
+
The raw data in this project are collected from eight open-source F* repositories on GitHib
|
100 |
+
1. [FStar](https://github.com/FStarLang/FStar): The F⋆ compiler itself, including its standard library and examples.
|
101 |
2. [Karamel](https://github.com/FStarLang/karamel): A transpiler from a subset of F⋆ called Low* to C, including libraries to work with a model of C types and control structures, e.g., for- and while-loops.
|
102 |
3. [EverParse](https://github.com/project-everest/everparse): A parser generator for binary formats, used in various large scale systems, e.g., the Windows kernel.
|
103 |
4. [HACL*](https://github.com/hacl-star/hacl-star): A library of verified cryptographic algorithms, including ValeCrypt, a library of verified assembly code, as well as EverCrypt, a cryptographic provider, including code deployed in Linux, Firefox, and Python.
|