Datasets:

Modalities:
Text
Formats:
parquet
Languages:
code
ArXiv:
Tags:
DOI:
License:
boomanaiden154 commited on
Commit
b9a6c18
1 Parent(s): 5334d06

Add more dataset size information

Browse files
Files changed (1) hide show
  1. README.md +33 -10
README.md CHANGED
@@ -39,15 +39,37 @@ task_ids: []
39
 
40
  |Release|Programming Languages|Description|
41
  |-|-|-|
42
- |v1.0| C/C++, Rust, Swift, Julia | Fine Tuning-scale dataset of 564GB of deduplicated LLVM IR |
43
 
44
  ### Dataset Summary
45
 
46
- ComPile contains over 500GB of permissively-licensed source code compiled to [LLVM](https://llvm.org) intermediate representation (IR) covering C/C++, Rust, Swift, and Julia.
 
47
  The dataset was created by hooking into LLVM code generation either through the language's package manager or the
48
  compiler directly to extract the dataset of intermediate representations from production grade programs using our
49
  [dataset collection utility for the LLVM compilation infrastructure](https://doi.org/10.5281/zenodo.10155761).
50
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
51
  ### Languages
52
 
53
  The dataset contains **5 programming languages** as of v1.0.
@@ -119,17 +141,17 @@ typically a link to a tar archive or git repository from which the project was b
119
  mapping to a specific package ecosystem that provides the source, such as Spack.
120
  - `language` (string): This column indicates the source language that the module was compiled from.
121
 
122
- ## Dataset Size
123
 
124
  | Langauge | Raw Size | License Constraints | Deduplicated + License Constraints |
125
  |----------|----------|---------------------|------------------------------------|
126
- | C/C++ | 124GB | 47GB | 31GB |
127
- | C | N/A | N/A | 3GB |
128
- | C++ | N/A | N/A | 28GB |
129
- | Julia | 201GB | 179GB | 153GB |
130
  | Swift | 8GB | 7GB | 7GB |
131
- | Rust | 656GB | 443GB | 373GB |
132
- | Total | 989GB | 676GB | 564GB |
133
 
134
  The raw size is the size obtained directly from building all the projects. The license constraints column
135
  shows the size per language after license information is taken into account. The last column shows the size
@@ -138,7 +160,8 @@ dataset.
138
 
139
  Note that the sizes displayed here are of the compressed bitcode representation rather
140
  than textual IR. We see an expansion ratio of 2-5x, averaging around 4x when converting
141
- from compressed bitcode to textual IR.
 
142
 
143
  ## Dataset Construction
144
 
 
39
 
40
  |Release|Programming Languages|Description|
41
  |-|-|-|
42
+ |v1.0| C/C++, Rust, Swift, Julia | Fine Tuning-scale dataset of 602GB of deduplicated LLVM (bitcode) IR |
43
 
44
  ### Dataset Summary
45
 
46
+ ComPile contains over 2.7TB of permissively-licensed source code compiled to (textual) [LLVM](https://llvm.org)
47
+ intermediate representation (IR) covering C/C++, Rust, Swift, and Julia.
48
  The dataset was created by hooking into LLVM code generation either through the language's package manager or the
49
  compiler directly to extract the dataset of intermediate representations from production grade programs using our
50
  [dataset collection utility for the LLVM compilation infrastructure](https://doi.org/10.5281/zenodo.10155761).
51
 
52
+ ### Dataset Size
53
+
54
+ The public release of ComPile contains over 2.7TB of textual LLVM-IR, which tokenizes into 1.3+T tokens using the Llama
55
+ tokenizer.
56
+
57
+ | Langauage | Bitcode Size | Textual IR Size | Llama Token Count | BPE Token Count (10k Vocab) | BPE Token Count (50k Vocab) |
58
+ |-----------|--------------|-----------------|-------------------|-----------------------------|-----------------------------|
59
+ | C | 2.47GB | 10.19GB | 5.31B | 0.91B | 0.58B |
60
+ | C++ | 28.87GB | 102.76GB | 46.75B | 11.20B | 6.27B |
61
+ | Julia | 164.16GB | 1088.39GB | 547.60B | 41.91B | 23.49B |
62
+ | Rust | 399.94GB | 1523.84GB | 735.90B | 137.37B | 90.01B |
63
+ | Swift | 6.95GB | 35.93GB | 19.78B | 3.36B | 1.75B |
64
+ | Total | 602.39GB | 2761.11GB | 1355.34B | 194.75B | 122.10B |
65
+
66
+ ComPile is distributed as bitcode, which is a compressed format that can be easily converted to and from the
67
+ textual representation of LLVM-IR. To collect token counts, we disassembled the bitcode to convert it into textual
68
+ IR and then ran a tokenizer over it. We used the standard Llama tokenizer and then ran fastBPE using a custom
69
+ vocabulary trained on a multi-GB sample of textual IR representativie of all languages in ComPile at two different
70
+ two different vocab sizes, particularly 10k and 50k. LLVM-IR is quite formulaic, so using custom vocabulary significantly
71
+ reduces the number of tokens generated.
72
+
73
  ### Languages
74
 
75
  The dataset contains **5 programming languages** as of v1.0.
 
141
  mapping to a specific package ecosystem that provides the source, such as Spack.
142
  - `language` (string): This column indicates the source language that the module was compiled from.
143
 
144
+ ## License Constraints and Deduplication
145
 
146
  | Langauge | Raw Size | License Constraints | Deduplicated + License Constraints |
147
  |----------|----------|---------------------|------------------------------------|
148
+ | C/C++ | 126GB | 46GB | 31GB |
149
+ | C | 16GB | N/A | 2GB |
150
+ | C++ | 109GB | N/A | 29GB |
151
+ | Julia | 201GB | 179GB | 164GB |
152
  | Swift | 8GB | 7GB | 7GB |
153
+ | Rust | 656GB | 443GB | 400GB |
154
+ | Total | 990GB | 675GB | 602GB |
155
 
156
  The raw size is the size obtained directly from building all the projects. The license constraints column
157
  shows the size per language after license information is taken into account. The last column shows the size
 
160
 
161
  Note that the sizes displayed here are of the compressed bitcode representation rather
162
  than textual IR. We see an expansion ratio of 2-5x, averaging around 4x when converting
163
+ from compressed bitcode to textual IR. Specific per-language numbers are available in the section
164
+ above on dataset size.
165
 
166
  ## Dataset Construction
167