HYUKI5 commited on
Commit
f091b44
·
verified ·
1 Parent(s): a73e473

nvidia/OpenCodeGeneticInstruct@9c95d36eb47c を複製し、カードを日本語で書く

Browse files
Files changed (33) hide show
  1. .gitattributes +0 -1
  2. README.md +398 -0
  3. mixtral-8x22b-instruct/train-00000-of-00018.parquet +3 -0
  4. mixtral-8x22b-instruct/train-00001-of-00018.parquet +3 -0
  5. mixtral-8x22b-instruct/train-00002-of-00018.parquet +3 -0
  6. mixtral-8x22b-instruct/train-00003-of-00018.parquet +3 -0
  7. mixtral-8x22b-instruct/train-00004-of-00018.parquet +3 -0
  8. mixtral-8x22b-instruct/train-00005-of-00018.parquet +3 -0
  9. mixtral-8x22b-instruct/train-00006-of-00018.parquet +3 -0
  10. mixtral-8x22b-instruct/train-00007-of-00018.parquet +3 -0
  11. mixtral-8x22b-instruct/train-00008-of-00018.parquet +3 -0
  12. mixtral-8x22b-instruct/train-00009-of-00018.parquet +3 -0
  13. mixtral-8x22b-instruct/train-00010-of-00018.parquet +3 -0
  14. mixtral-8x22b-instruct/train-00011-of-00018.parquet +3 -0
  15. mixtral-8x22b-instruct/train-00012-of-00018.parquet +3 -0
  16. mixtral-8x22b-instruct/train-00013-of-00018.parquet +3 -0
  17. mixtral-8x22b-instruct/train-00014-of-00018.parquet +3 -0
  18. mixtral-8x22b-instruct/train-00015-of-00018.parquet +3 -0
  19. mixtral-8x22b-instruct/train-00016-of-00018.parquet +3 -0
  20. mixtral-8x22b-instruct/train-00017-of-00018.parquet +3 -0
  21. qwen2.5-32b-instruct/train-00000-of-00013.parquet +3 -0
  22. qwen2.5-32b-instruct/train-00001-of-00013.parquet +3 -0
  23. qwen2.5-32b-instruct/train-00002-of-00013.parquet +3 -0
  24. qwen2.5-32b-instruct/train-00003-of-00013.parquet +3 -0
  25. qwen2.5-32b-instruct/train-00004-of-00013.parquet +3 -0
  26. qwen2.5-32b-instruct/train-00005-of-00013.parquet +3 -0
  27. qwen2.5-32b-instruct/train-00006-of-00013.parquet +3 -0
  28. qwen2.5-32b-instruct/train-00007-of-00013.parquet +3 -0
  29. qwen2.5-32b-instruct/train-00008-of-00013.parquet +3 -0
  30. qwen2.5-32b-instruct/train-00009-of-00013.parquet +3 -0
  31. qwen2.5-32b-instruct/train-00010-of-00013.parquet +3 -0
  32. qwen2.5-32b-instruct/train-00011-of-00013.parquet +3 -0
  33. qwen2.5-32b-instruct/train-00012-of-00013.parquet +3 -0
.gitattributes CHANGED
@@ -1,6 +1,5 @@
1
  *.7z filter=lfs diff=lfs merge=lfs -text
2
  *.arrow filter=lfs diff=lfs merge=lfs -text
3
- *.avro filter=lfs diff=lfs merge=lfs -text
4
  *.bin filter=lfs diff=lfs merge=lfs -text
5
  *.bz2 filter=lfs diff=lfs merge=lfs -text
6
  *.ckpt filter=lfs diff=lfs merge=lfs -text
 
1
  *.7z filter=lfs diff=lfs merge=lfs -text
2
  *.arrow filter=lfs diff=lfs merge=lfs -text
 
3
  *.bin filter=lfs diff=lfs merge=lfs -text
4
  *.bz2 filter=lfs diff=lfs merge=lfs -text
5
  *.ckpt filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,398 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ pretty_name: nvidia/OpenCodeGeneticInstruct (mirror)
4
+ source_datasets:
5
+ - nvidia/OpenCodeGeneticInstruct
6
+ tags:
7
+ - laya
8
+ - laya-finetune
9
+ - mirror
10
+ - domain-code
11
+ - use-permissive
12
+ - genre-instruction
13
+ - label-text
14
+ - annot-llm
15
+ - lang-en
16
+ language:
17
+ - en
18
+ size_categories:
19
+ - 10M<n<100M
20
+ configs:
21
+ - config_name: mixtral-8x22b-instruct
22
+ data_files:
23
+ - split: train
24
+ path: mixtral-8x22b-instruct/train-*
25
+ - config_name: qwen2.5-32b-instruct
26
+ data_files:
27
+ - split: train
28
+ path: qwen2.5-32b-instruct/train-*
29
+ dataset_info:
30
+ - config_name: mixtral-8x22b-instruct
31
+ features:
32
+ - name: id
33
+ dtype: string
34
+ - name: input
35
+ dtype: string
36
+ - name: solution
37
+ dtype: string
38
+ - name: output
39
+ dtype: string
40
+ - name: last_operation
41
+ dtype: string
42
+ - name: generation_model
43
+ dtype: string
44
+ splits:
45
+ - name: train
46
+ num_bytes: 34330192247
47
+ num_examples: 7593134
48
+ download_size: 16733055803
49
+ dataset_size: 34330192247
50
+ - config_name: qwen2.5-32b-instruct
51
+ features:
52
+ - name: id
53
+ dtype: string
54
+ - name: input
55
+ dtype: string
56
+ - name: solution
57
+ dtype: string
58
+ - name: output
59
+ dtype: string
60
+ - name: last_operation
61
+ dtype: string
62
+ - name: generation_model
63
+ dtype: string
64
+ splits:
65
+ - name: train
66
+ num_bytes: 24957035531
67
+ num_examples: 7500000
68
+ download_size: 10135336305
69
+ dataset_size: 24957035531
70
+ ---
71
+
72
+ # nvidia/OpenCodeGeneticInstruct(複製)
73
+
74
+ ## データの内容
75
+
76
+ Genetic-Instruct の手法で合成した、Python のコーディング指示と解答のデータ。1 行が 1 件の質問と応答で、質問 `input`、応答全体 `output`、コードだけの `solution`、生成に使った操作 `last_operation`(mutation か crossover)と生成モデル `generation_model` を持つ。種になる問題は Tiger-Leetcode の 512 件で、生成は Mixtral-8x22B-Instruct と Qwen2.5-32B-Instruct が行った。
77
+
78
+ ## データの概要
79
+
80
+ | 項目 | 値 |
81
+ |---|---|
82
+ | 行数 | 15,093,134(2 構成・2 分割の合計) |
83
+ | 列数 | 6 |
84
+ | 容量 | 26.87 GB(ファイル 33 本) |
85
+ | ファイルの形式 | parquet 31 |
86
+ | 言語 | en |
87
+ | 公開 | public |
88
+
89
+ ### 構成と分割
90
+
91
+ | 構成(config) | 分割(split)ごとの行数 | 列数 |
92
+ |---|---|---:|
93
+ | `mixtral-8x22b-instruct` | `train` 7,593,134 | 6 |
94
+ | `qwen2.5-32b-instruct` | `train` 7,500,000 | 6 |
95
+
96
+ ### 列(構成 `mixtral-8x22b-instruct`・分割 `train`。どの構成も同じ列)
97
+
98
+ | 列 | 型 | 先頭の行の値 |
99
+ |---|---|---|
100
+ | `id` | `string` | `617d30a7f75a4777b1def52c5bf3dcc8` |
101
+ | `input` | `string` | `Implement the __<method/function>__ pattern, aft…` |
102
+ | `solution` | `string` | `import threading import time class Calculator: …` |
103
+ | `output` | `string` | `The described problem can be solved by using the…` |
104
+ | `last_operation` | `string` | `mutation` |
105
+ | `generation_model` | `string` | `Mixtral-8x22B-Instruct` |
106
+
107
+ 行数と列は HF のデータセットビューアの集計(元のデータセットの現在の版で数えたもの)。全データセットの一覧は [HYUKI5/dataset-catalog](https://huggingface.co/datasets/HYUKI5/dataset-catalog) にある。
108
+
109
+ ## 複製について
110
+
111
+ [`nvidia/OpenCodeGeneticInstruct`](https://huggingface.co/datasets/nvidia/OpenCodeGeneticInstruct) をリビジョン `9c95d36eb47c3d8482066282112ad3d057b9ed51` の時点でそのまま複製したもの。中身は変えていない。判断モデル Laya(`convaiinnovations/laya`)の微調整に使う候補として、調査メモ(`Laya 学習用データセット調査`)に載せたデータを集めている。
112
+
113
+ ## 出典
114
+
115
+ - 元のデータセット: https://huggingface.co/datasets/nvidia/OpenCodeGeneticInstruct
116
+ - 複製したリビジョン: `9c95d36eb47c3d8482066282112ad3d057b9ed51`(2025-05-23 00:13:04+00:00)
117
+ - 元のライセンス表記: `cc-by-4.0`
118
+
119
+ ## 論文
120
+
121
+ - https://arxiv.org/abs/2407.21077
122
+
123
+ ## アノテーションの方法
124
+
125
+ **データセットの作成期間**
126
+
127
+ 2025 年 1 月〜2025 年 3 月
128
+
129
+ ## 微調整に使うときの注意
130
+
131
+ - 調査メモの台帳での区分: A(許諾型のライセンスだけで、疑義の注記が無い)。区分は記載の文言で機械的に付けたもので、法的な判定ではない。学習に入れる前に元のカードとライセンスの原文で確かめる。
132
+ - Laya の入力は 2,048 token まで(ModernBERT-large / mmBERT)。長い本文は先頭から黙って切られるので、切れる行は事前に除くか、要約・切り出しをしてから入れる。
133
+ - 分割は記事・文書単位で行い、同じ本文から作った別の問いを学習側と較正側に分けない。温度と閾値は学習に使っていない分割で別々に当てる。
134
+
135
+ ## 調査メモの記述
136
+
137
+ | メモ | 節 | ライセンス(記載) | 確認 | 注記 |
138
+ |---|---|---|---|---|
139
+ | `nvidia_nemotron.md` | 付録. 取得した nvidia 組織のデータセット全件の簡易一覧 (L401) | cc-by-4.0 | 確認済 | 確認印なし(各ファイルの凡例により確認済扱い) / 原文注記: × 数学・コード・科学・計測 |
140
+
141
+ ## 元のカード(日本語訳)
142
+
143
+ 以下は元のデータセットのカードの本文を日本語に訳したもの(機械翻訳)。訳に疑問があるときは、末尾の原文を確かめる。
144
+
145
+ ---
146
+
147
+ # OpenCodeGeneticInstruct: コードを生成する LLM の能力を高めるための、大規模なコーディング指示のデータセット
148
+
149
+ ## データの概要
150
+
151
+ OpenCodeGeneticInstruct は、Genetic-Instruct [1] の手法で合成的に生成された、Python のコーディング指示を 1,500 万件以上含む。このデータセットは、LLM のコード生成能力を高めるための教師あり微調整(SFT)に使える。各サンプルは、コーディングの質問または指示と、その答えからなる。答えには Python によるコードの解答が含まれる。
152
+
153
+ - [論文](https://arxiv.org/abs/2407.21077) - OpenCodeGeneticInstruct の方法論と技術的詳細を確認できる。
154
+ - [Github リポジトリ](https://github.com/NVIDIA/NeMo-Skills) - SFT の実験に使われた完全なパイプラインを参照できる。
155
+
156
+ このデータセットは商用・非商用のどちらでも利用できる。
157
+
158
+ ## データの分布
159
+
160
+ - このデータセットの生成では、初期集団として Tiger-Leetcode [TigerResearch. 2023. Tigerbot kaggle leetcode solutions dataset (english) - 2k. https://huggingface.co/datasets/TigerResearch/tigerbot-kaggle-leetcodesolutions-en-2k.] のコレクションから 512 件のサンプルを使った。
161
+ - 2 つの生成モデル、Mixtral-8x22B-Instruct と Qwen2.5-32B-Instruct を使ってサンプルを生成した。
162
+
163
+ | Generation Model | # Seed | # Sample |
164
+ |:--------------------------|:-----------|:-----------|
165
+ | Mixtral-8x22B-Instruct | 512 | 7,593,134 |
166
+ | Qwen2.5-32B-Instruct | 512 | 7,000,000 |
167
+ | 合計 | 512 | 15,093,134 |
168
+
169
+ ## データセットの提供者
170
+
171
+ NVIDIA Corporation
172
+
173
+
174
+ ## データセットの作成期間
175
+
176
+ 2025 年 1 月〜2025 年 3 月
177
+
178
+
179
+ ## ライセンス・利用条件
180
+
181
+ 準拠規約: このデータセットは [Creative Commons Attribution 4.0 International License(CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/legalcode) のもとで提供される。
182
+
183
+ **データの開発者:** NVIDIA
184
+
185
+ ### ユースケース: <br>
186
+ LLM をコード生成に特化させるために LLM を訓練する開発者。 <br>
187
+
188
+ ### 公開日: <br>
189
+ 05/06/2025 <br>
190
+
191
+ ## 想定される用途
192
+
193
+ OpenCodeInstruct データセットは、コミュニティがオープンなモデルを改善し続けるために使われることを想定している。このデータは自由に使ってモデルを訓練してよい。**ただし、利用者が使うデータセットごとに、そのライセンスが目的に適うかは利用者の責任で確かめること**。
194
+
195
+
196
+ ## データセットの特性
197
+ ** データ収集方法<br>
198
+ * [ハイブリッド: 自動, 合成] <br>
199
+
200
+ ** ラベル付けの方法<be>
201
+ * [ハイブリッド: 自動, 合成] <br>
202
+
203
+
204
+ ## データの形式
205
+ |Field|Type|Description|
206
+ |:---|:---|:---|
207
+ |id|string|質問ごとの一意な ID|
208
+ |input|string|競技プログラミングの入力となる質問|
209
+ |output|string|モデルの応答。|
210
+ |solution|string|モデルの応答のうち、コードの部分だけ。|
211
+ |last_operation|string|この質問を生成するのに使われた操作(mutation、crossover)。|
212
+ |generation_model|string|この質問や応答を生成するのに使われたモデル(Mixtral-8x22B-Instruct、Qwen2.5-32B-Instruct)。|
213
+
214
+
215
+ ## 使い方
216
+
217
+ 次のコードでデータセットを読み込める。
218
+
219
+ ```python
220
+ from datasets import load_dataset
221
+
222
+ ds_mixtral = load_dataset("nvidia/OpenCodeGeneticInstruct",
223
+ name="mixtral-8x22b-instruct", split="train")
224
+ print(ds_mixtral)
225
+
226
+ ## output:
227
+ ## Dataset({
228
+ ## features: ['id', 'input', 'solution', 'output', 'operation', 'model'],
229
+ ## num_rows: 7593134
230
+ ## })
231
+
232
+ ds_qwen = load_dataset("nvidia/OpenCodeGeneticInstruct",
233
+ name="qwen2.5-32b-instruct", split="train")
234
+ print(ds_qwen)
235
+
236
+ ## output:
237
+ ## Dataset({
238
+ ## features: ['id', 'input', 'solution', 'output', 'operation', 'model'],
239
+ ## num_rows: 7500000
240
+ ##})
241
+ ```
242
+ ## データ量
243
+
244
+ - レコード数 - 1,500 万件の、コーディングの質問と答えのペア。
245
+ - ダウンロードサイズ - 25 GB
246
+
247
+
248
+ ## 倫理的配慮:
249
+
250
+ NVIDIA は、信頼できる AI は共同の責任であると考えており、多様な AI 応用の開発を可能にする方針と慣行を整えている。ダウンロードして利用規約に従って使う場合、開発者は社内のモデルチームと協力して、このモデルが関連する業界と用途の要件を満たすこと、そして想定外の製品上の誤用に対処��ることを確かめるべきである。
251
+
252
+ セキュリティの脆弱性や NVIDIA の AI に関する懸念は [こちら](https://www.nvidia.com/en-us/support/submit-security-vulnerability/) から報告してほしい。
253
+
254
+
255
+ ## 引用
256
+
257
+ データが有用だった場合は、次の文献を引用してほしい。
258
+ ```
259
+ [1] @article{majumdar2025geneticinstructscalingsynthetic,
260
+ title = {Genetic Instruct: Scaling up Synthetic Generation of Coding Instructions for Large Language Models},
261
+ author = {Somshubra Majumdar, Vahid Noroozi, Mehrzad Samadi, Sean Narenthiran, Aleksander Ficek, Wasi Uddin Ahmad, Jocelyn Huang, Jagadeesh Balam, Boris Ginsburg},
262
+ year={2025},
263
+ eprint={2407.21077},
264
+ archivePrefix={arXiv},
265
+ primaryClass={cs.CL},
266
+ url={https://arxiv.org/abs/2407.21077},
267
+ }
268
+ ```
269
+
270
+ <details>
271
+ <summary>原文(英語)</summary>
272
+
273
+ # OpenCodeGeneticInstruct: A large-scale dataset of coding instructions for improving the code generation capabilities of LLMs
274
+
275
+ ## Data Overview
276
+
277
+ OpenCodeGeneticInstruct comprises more than 15M coding instructions in python which is generated synthetically with the Genetic-Instruct [1] approach.
278
+ This dataset can be used for supervised fine-tuning (SFT) of LLMs to improve their code genearation capability.
279
+ Each sample includes a coding question/instruction and its corrsponding answer. The answer contains the coding solution in Python.
280
+
281
+ - [Paper](https://arxiv.org/abs/2407.21077) - Discover the methodology and technical details behind OpenCodeGeneticInstruct.
282
+ - [Github Repo](https://github.com/NVIDIA/NeMo-Skills) - Access the complete pipeline used to perform the SFT experiments.
283
+
284
+ This dataset can be used for commercial/non-commercial use.
285
+
286
+ ## Data distribution
287
+
288
+ - We use 512 samples from the Tiger-Leetcode [TigerResearch. 2023. Tigerbot kaggle leetcode solutions dataset (english) - 2k. https://huggingface.co/datasets/TigerResearch/tigerbot-kaggle-leetcodesolutions-en-2k.] collection as the initial population to generate this dataset.
289
+ - Two genearation models, Mixtral-8x22B-Instruct and Qwen2.5-32B-Instruct, are used to generate the samples.
290
+
291
+ | Generation Model | # Seed | # Sample |
292
+ |:--------------------------|:-----------|:-----------|
293
+ | Mixtral-8x22B-Instruct | 512 | 7,593,134 |
294
+ | Qwen2.5-32B-Instruct | 512 | 7,000,000 |
295
+ | Total | 512 | 15,093,134 |
296
+
297
+ ## Dataset Owner(s)
298
+ NVIDIA Corporation
299
+
300
+
301
+ ## Dataset Creation Date
302
+ January 2025 - March 2025
303
+
304
+
305
+ ## License/Terms of Use
306
+
307
+ GOVERNING TERMS: This dataset is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0) available at https://creativecommons.org/licenses/by/4.0/legalcode.
308
+
309
+ **Data Developer:** NVIDIA
310
+
311
+ ### Use Case: <br>
312
+ Developers training LLMs to specialize LLMs in code generation. <br>
313
+
314
+ ### Release Date: <br>
315
+ 05/06/2025 <br>
316
+
317
+ ## Intended Usage
318
+
319
+ The OpenCodeInstruct Dataset is intended to be used by the community to continue to improve open models. The data may be freely used to train models. **However, for
320
+ each dataset a user elects to use, the user is responsible for checking if the dataset license is fit for the intended purpose**.
321
+
322
+
323
+ ## Dataset Characterization
324
+
325
+ ** Data Collection Method<br>
326
+ * [Hybrid: Automated, Synthetic] <br>
327
+
328
+ ** Labeling Method<be>
329
+ * [Hybrid: Automated, Synthetic] <br>
330
+
331
+
332
+ ## Dataset Format
333
+ |Field|Type|Description|
334
+ |:---|:---|:---|
335
+ |id|string|A unique id for each question|
336
+ |input|string|The input competitive programming question |
337
+ |output|string|Model's response.|
338
+ |solution|string|Only the code portion of Model's response.|
339
+ |last_operation|string|Operation used to generate this question (mutation, crossover).|
340
+ |generation_model|string|Model used to generate this question/response (Mixtral-8x22B-Instruct, Qwen2.5-32B-Instruct).|
341
+
342
+
343
+ ## How to use it
344
+
345
+ You can load the dataset with the following lines of code.
346
+
347
+ ```python
348
+ from datasets import load_dataset
349
+
350
+ ds_mixtral = load_dataset("nvidia/OpenCodeGeneticInstruct",
351
+ name="mixtral-8x22b-instruct", split="train")
352
+ print(ds_mixtral)
353
+
354
+ ## output:
355
+ ## Dataset({
356
+ ## features: ['id', 'input', 'solution', 'output', 'operation', 'model'],
357
+ ## num_rows: 7593134
358
+ ## })
359
+
360
+ ds_qwen = load_dataset("nvidia/OpenCodeGeneticInstruct",
361
+ name="qwen2.5-32b-instruct", split="train")
362
+ print(ds_qwen)
363
+
364
+ ## output:
365
+ ## Dataset({
366
+ ## features: ['id', 'input', 'solution', 'output', 'operation', 'model'],
367
+ ## num_rows: 7500000
368
+ ##})
369
+ ```
370
+ ## Dataset Quantification
371
+
372
+ - Record Count - 15 million coding question-answer pairs.
373
+ - Download Size - 25 GB
374
+
375
+
376
+ ## Ethical Considerations:
377
+
378
+ NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.
379
+
380
+ Please report security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/).
381
+
382
+
383
+ ## Citation
384
+
385
+ If you find the data useful, please cite:
386
+ ```
387
+ [1] @article{majumdar2025geneticinstructscalingsynthetic,
388
+ title = {Genetic Instruct: Scaling up Synthetic Generation of Coding Instructions for Large Language Models},
389
+ author = {Somshubra Majumdar, Vahid Noroozi, Mehrzad Samadi, Sean Narenthiran, Aleksander Ficek, Wasi Uddin Ahmad, Jocelyn Huang, Jagadeesh Balam, Boris Ginsburg},
390
+ year={2025},
391
+ eprint={2407.21077},
392
+ archivePrefix={arXiv},
393
+ primaryClass={cs.CL},
394
+ url={https://arxiv.org/abs/2407.21077},
395
+ }
396
+ ```
397
+
398
+ </details>
mixtral-8x22b-instruct/train-00000-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42045ae7ea1b442ba7ba65b8191a1c18cd9f91195c0610bea654fa9fe4b5078c
3
+ size 822654102
mixtral-8x22b-instruct/train-00001-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7644353e997a30047ef14998b16b6917bb962c4f0fdb38696b96143fc80e7ac5
3
+ size 824445856
mixtral-8x22b-instruct/train-00002-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:60927f5b143af638973de50b609c96cb86f4a6dc66f62f21aa023fce40028110
3
+ size 829391100
mixtral-8x22b-instruct/train-00003-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e94c18f2dd6d9eb629b43ae94e96d18ceb13aa6d6bba19f700bdbe6a2d82ec7
3
+ size 851590363
mixtral-8x22b-instruct/train-00004-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a9713035619dab915e7553379ea927b9b12bcfb31b123dee4f5aaa818651f174
3
+ size 882768948
mixtral-8x22b-instruct/train-00005-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eab0e63468f5d0045d6535a471bbed3c5f7954cb88f37c3fbc3a72235ea28e8f
3
+ size 883230798
mixtral-8x22b-instruct/train-00006-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ccd1061aacd2d20810dc1131f9d18b1e716e31fa699667def617b9da88b32165
3
+ size 884985195
mixtral-8x22b-instruct/train-00007-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:23f926a988e3356a075fa64874f3e3777e53db0f51a4d829adf1f36591d651d7
3
+ size 934883722
mixtral-8x22b-instruct/train-00008-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07ae4e42a63674de761e411e9500060b65bb56a1d023856281a4e47c0ef76442
3
+ size 936926641
mixtral-8x22b-instruct/train-00009-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc6d46f9b9e88d49f429b36ec3f4d7dd27e48de4f7cadb8f019c7d030aff2601
3
+ size 932549810
mixtral-8x22b-instruct/train-00010-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4d64f345c2a0090d14fb72b5ed0e9ccbb8ed5f7ebaaca4fe2839fe68b606da2
3
+ size 948067705
mixtral-8x22b-instruct/train-00011-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9bb78b07658cb1eb5d26b7263a634cc2c38575e02984965683da22eb0f57e589
3
+ size 981979339
mixtral-8x22b-instruct/train-00012-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa21acb1dafa7981285bbd8ceefc1b15c55c15ae19db69a5b57d2448f0878646
3
+ size 980099732
mixtral-8x22b-instruct/train-00013-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1026b508ec0f00f8cd79b422224ad9f80cf8618db6613ae62e787040576e2519
3
+ size 980458943
mixtral-8x22b-instruct/train-00014-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a19e8cc8eb6c131c53186ecad72553ca3aedcff83f835ef18b5b74d212b35bfb
3
+ size 1005576992
mixtral-8x22b-instruct/train-00015-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:792ffc69c204a47256ba87bf33f7c46ab2d0d08056364fbca8bf08e89a685896
3
+ size 1015910993
mixtral-8x22b-instruct/train-00016-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be98b66d45d6ac6c67f332b9fc6b809bb0c0b8259eba7f1029f3c2d5f0281710
3
+ size 1017734445
mixtral-8x22b-instruct/train-00017-of-00018.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8be8f6088aa47ff930700391d6ac42f04c0bd76f5ef6ce5672dbed7f3e6d19bc
3
+ size 1019801119
qwen2.5-32b-instruct/train-00000-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10d1b66f270b0d473f60bef0ad6ecebf6759788b0e5ffbf74ee21c7519307785
3
+ size 851935274
qwen2.5-32b-instruct/train-00001-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7e8e7f72009cde6d56d0b94a4769f6030f9c603ba0accb15aa075bc32402738
3
+ size 853415546
qwen2.5-32b-instruct/train-00002-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:771e92578f43c98eb09d1097227f05bb1b16f5337d002910ed989189c3262c3e
3
+ size 837650766
qwen2.5-32b-instruct/train-00003-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1c6baaf7522765c4f2302640c2abb0d60c4ee98515896a35e12393d00263317
3
+ size 814805222
qwen2.5-32b-instruct/train-00004-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df97765f4aaada45cb2055a8e68d529cd49301ecacd0b49be7601bf8a511bec4
3
+ size 815331217
qwen2.5-32b-instruct/train-00005-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9660656537d95b149c58b1aea65bf4bf40fb835ab01254a45157e64866aa6307
3
+ size 785738160
qwen2.5-32b-instruct/train-00006-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0180b570623b6bb4d97136f290cc73492e74760ab7b617efecdfdbdabbce193e
3
+ size 777830695
qwen2.5-32b-instruct/train-00007-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:75161c76d5fe2b589cbcb948c831c72a4e8c3f22887f1055aa38321e18df9cf6
3
+ size 771506427
qwen2.5-32b-instruct/train-00008-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1156a57ff38b8831f2d0996d40ad9b3e199c9440e5b956e4bc46df0311d17748
3
+ size 741609418
qwen2.5-32b-instruct/train-00009-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0fd839b8c3ffa606b387e5b87a341c788d830957e459f6743bd91043792f95d0
3
+ size 742339566
qwen2.5-32b-instruct/train-00010-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98f5cbd851b1ef260fa6ee7f338cb312fda6d8113362f2350d8742453e3868df
3
+ size 723208518
qwen2.5-32b-instruct/train-00011-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4de8eae1084b3b463007196887e76ba9c413f419d97602561c1bae08ba141481
3
+ size 709988627
qwen2.5-32b-instruct/train-00012-of-00013.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d598defdc24e0b8dc4cf2c8f163a1eac7f977f2dd3be8f873d712ddcad9d1500
3
+ size 709976869