Update README.md
Browse files
README.md
CHANGED
|
@@ -36,14 +36,13 @@ Read more details about Sailor2 at https://sailorllm.github.io/blog/sailor2.
|
|
| 36 |
<b><font size="+1">π Sailor2 Pre-training Dataset </font></b>
|
| 37 |
</summary>
|
| 38 |
|
| 39 |
-
- [Sailor2-pretrain-data-stage1](https://huggingface.co/datasets/sailor2/sailor2-pretrain-data-stage1)
|
| 40 |
-
- [Sailor2-pretrain-data-stage2](https://huggingface.co/datasets/sailor2/sailor2-pretrain-data-stage2)
|
| 41 |
-
- [sea-commoncrawl](https://huggingface.co/datasets/sailor2/sea-commoncrawl)
|
| 42 |
-
- [sea-internet](https://huggingface.co/datasets/sailor2/sea-internet)
|
| 43 |
-
- [sea-
|
| 44 |
-
- [sea-
|
| 45 |
-
- [sea-
|
| 46 |
-
- [sea-commoncrawl-high-quality](https://huggingface.co/datasets/sailor2/sea-commoncrawl-high-quality)
|
| 47 |
|
| 48 |
</details>
|
| 49 |
|
|
@@ -54,11 +53,35 @@ Read more details about Sailor2 at https://sailorllm.github.io/blog/sailor2.
|
|
| 54 |
<b><font size="+1">π Sailor2 Post-training Dataset </font></b>
|
| 55 |
</summary>
|
| 56 |
|
| 57 |
-
- [sailor2-sft-stage1](https://huggingface.co/datasets/sailor2/sailor2-sft-stage1)
|
| 58 |
-
- [sailor2-sft-stage2](https://huggingface.co/datasets/sailor2/sailor2-sft-stage2)
|
| 59 |
-
- [sea-ultrafeedback](https://huggingface.co/datasets/sailor2/sea-ultrafeedback)
|
| 60 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
|
|
|
|
|
|
|
| 62 |
</details>
|
| 63 |
|
| 64 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
<b><font size="+1">π Sailor2 Pre-training Dataset </font></b>
|
| 37 |
</summary>
|
| 38 |
|
| 39 |
+
- [Sailor2-pretrain-data-stage1](https://huggingface.co/datasets/sailor2/sailor2-pretrain-data-stage1): 500B high quality data for model training
|
| 40 |
+
- [Sailor2-pretrain-data-stage2](https://huggingface.co/datasets/sailor2/sailor2-pretrain-data-stage2): 50B extra high quality data for model annealing
|
| 41 |
+
- [sea-commoncrawl](https://huggingface.co/datasets/sailor2/sea-commoncrawl): Cleaned and deduplicated commoncrawl
|
| 42 |
+
- [sea-internet](https://huggingface.co/datasets/sailor2/sea-internet): Cleaned multilingual data from Internet Archive
|
| 43 |
+
- [sea-pdf-text](https://huggingface.co/datasets/sailor2/sea-pdf-text): Cleaned pdf data
|
| 44 |
+
- [sea-syntheitc](https://huggingface.co/datasets/sailor2/sea-syntheitc): Translation dataset from Cosmopedia across multiple languages
|
| 45 |
+
- [sea-commoncrawl-high-quality](https://huggingface.co/datasets/sailor2/sea-commoncrawl-high-quality): extra cleaned and deduplicated commoncrawl
|
|
|
|
| 46 |
|
| 47 |
</details>
|
| 48 |
|
|
|
|
| 53 |
<b><font size="+1">π Sailor2 Post-training Dataset </font></b>
|
| 54 |
</summary>
|
| 55 |
|
| 56 |
+
- [sailor2-sft-stage1](https://huggingface.co/datasets/sailor2/sailor2-sft-stage1): Medium-Quality Instruction tuning dataset, supports English, Chinese and 15 SEA languages.
|
| 57 |
+
- [sailor2-sft-stage2](https://huggingface.co/datasets/sailor2/sailor2-sft-stage2): High-Quality Instruction tuning dataset, supports English, Chinese and 15 SEA languages.
|
| 58 |
+
- [sea-ultrafeedback](https://huggingface.co/datasets/sailor2/sea-ultrafeedback): Preference optimization dataset, supports English, Chinese and 17 SEA languages.
|
| 59 |
+
|
| 60 |
+
</details>
|
| 61 |
+
|
| 62 |
+
---
|
| 63 |
+
|
| 64 |
+
<details>
|
| 65 |
+
<summary>
|
| 66 |
+
<b><font size="+1">π§ Sailor2 Evaluation Dataset </font></b>
|
| 67 |
+
</summary>
|
| 68 |
|
| 69 |
+
- [sea-wildbench](https://huggingface.co/datasets/sailor2/sea-wildbench): Chat model evaluation, supports 8 SEA languages.
|
| 70 |
+
|
| 71 |
</details>
|
| 72 |
|
| 73 |
---
|
| 74 |
+
|
| 75 |
+
<details>
|
| 76 |
+
<summary>
|
| 77 |
+
<b><font size="+1">π» Sailor2 Codebase </font></b>
|
| 78 |
+
</summary>
|
| 79 |
+
|
| 80 |
+
- [SailCraft Code](https://github.com/sail-sg/sailcraft): Data cleaning
|
| 81 |
+
- [Regmix Code](https://github.com/sail-sg/regmix): Data mixture
|
| 82 |
+
- [SailCompass Code](https://huggingface.co/datasets/sailor2/sailor2-sft-stage1): Few-shot evaluation
|
| 83 |
+
- [Megatron Code](https://github.com/sail-sg/Megatron-Sailor2): Pretraining-training
|
| 84 |
+
- [OAT Code](https://github.com/sail-sg/oat): Post-training
|
| 85 |
+
|
| 86 |
+
</details>
|
| 87 |
+
|