Upload README.md
Browse files
README.md
CHANGED
|
@@ -1,16 +1,16 @@
|
|
| 1 |
# Introduction
|
| 2 |
|
| 3 |
-
DeepSeek-R1-FlagOS-Metax-BF16 provides an all-in-one deployment solution, enabling execution of DeepSeek-R1 on Metax GPUs. As the first-generation release for the Metax
|
| 4 |
|
| 5 |
1. Comprehensive Integration:
|
| 6 |
- Integrated with FlagScale (https://github.com/FlagOpen/FlagScale).
|
| 7 |
- Open-source inference execution code, preconfigured with all necessary software and hardware settings.
|
| 8 |
- Verified model files, available on Hugging Face ([Model Link](https://huggingface.co/FlagRelease/DeepSeek-R1-FlagOS-Metax-BF16)).
|
| 9 |
-
- Pre-built Docker image for rapid deployment on Metax
|
| 10 |
2. High-Precision BF16 Checkpoints:
|
| 11 |
- BF16 checkpoints dequantized from the official DeepSeek-R1 FP8 model to ensure enhanced inference accuracy and performance.
|
| 12 |
3. Consistency Validation:
|
| 13 |
-
- Evaluation tests verifying consistency of results between NVIDIA H100 and Metax
|
| 14 |
|
| 15 |
# Technical Summary
|
| 16 |
|
|
@@ -141,12 +141,11 @@ We warmly welcome global developers to join us:
|
|
| 141 |
Scan the QR code below to add our WeChat group
|
| 142 |
send "FlagRelease"
|
| 143 |
|
| 144 |
-
.
|
| 150 |
|
| 151 |
<p style="color: lightgrey;">如果您是本模型的贡献者，我们邀请您根据<a href="https://modelscope.cn/docs/ModelScope%E6%A8%A1%E5%9E%8B%E6%8E%A5%E5%85%A5%E6%B5%81%E7%A8%8B%E6%A6%82%E8%A7%88" style="color: lightgrey; text-decoration: underline;">模型贡献文档</a>，及时完善模型卡片内容。</p>
|
| 152 |
|
|
|
|
| 1 |
# Introduction
|
| 2 |
|
| 3 |
+
DeepSeek-R1-FlagOS-Metax-BF16 provides an all-in-one deployment solution, enabling execution of DeepSeek-R1 on Metax GPUs. As the first-generation release for the Metax series, this package delivers three key features:
|
| 4 |
|
| 5 |
1. Comprehensive Integration:
|
| 6 |
- Integrated with FlagScale (https://github.com/FlagOpen/FlagScale).
|
| 7 |
- Open-source inference execution code, preconfigured with all necessary software and hardware settings.
|
| 8 |
- Verified model files, available on Hugging Face ([Model Link](https://huggingface.co/FlagRelease/DeepSeek-R1-FlagOS-Metax-BF16)).
|
| 9 |
+
- Pre-built Docker image for rapid deployment on Metax.
|
| 10 |
2. High-Precision BF16 Checkpoints:
|
| 11 |
- BF16 checkpoints dequantized from the official DeepSeek-R1 FP8 model to ensure enhanced inference accuracy and performance.
|
| 12 |
3. Consistency Validation:
|
| 13 |
+
- Evaluation tests verifying consistency of results between NVIDIA H100 and Metax.
|
| 14 |
|
| 15 |
# Technical Summary
|
| 16 |
|
|
|
|
| 141 |
Scan the QR code below to add our WeChat group
|
| 142 |
send "FlagRelease"
|
| 143 |
|
| 144 |
+
|
| 145 |
|
| 146 |
# License
|
| 147 |
|
| 148 |
+
The weights of this model are based on deepseek-ai/DeepSeek-R1 and are open-sourced under the Apache 2.0 License: https://www.apache.org/licenses/LICENSE-2.0.txt.
|
|
|
|
| 149 |
|
| 150 |
<p style="color: lightgrey;">如果您是本模型的贡献者，我们邀请您根据<a href="https://modelscope.cn/docs/ModelScope%E6%A8%A1%E5%9E%8B%E6%8E%A5%E5%85%A5%E6%B5%81%E7%A8%8B%E6%A6%82%E8%A7%88" style="color: lightgrey; text-decoration: underline;">模型贡献文档</a>，及时完善模型卡片内容。</p>
|
| 151 |
|