github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-08-11 12:51:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

[misc] Add dist optim to doc sidebar (#5806)

Browse Source 
* add to sidebar

* fix chinese
This commit is contained in:
Edenzzzz 2024-06-18 13:52:47 +08:00 committed by GitHub
parent 8795bb2e80
commit 7f9ec599be
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 21 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
docs/sidebars.json
View File

@ -56,6 +56,7 @@
"features/pipeline_parallel", "features/pipeline_parallel",
"features/nvme_offload", "features/nvme_offload",
"features/lazy_init", "features/lazy_init",
"features/distributed_optimizers",
"features/cluster_utils" "features/cluster_utils"
] ]
}, },

13
docs/source/en/features/distributed_optimizers.md
View File

@ -14,12 +14,6 @@ Apart from the widely adopted Adam and SGD, many modern optimizers require layer
## Optimizers ## Optimizers
Adafactor is a first-order Adam variant using Non-negative Matrix Factorization(NMF) to reduce memory footprint. CAME improves by introducting a confidence matrix to correct NMF. GaLore further reduces memory by projecting gradients into a low-rank space and 8-bit block-wise quantization. Lamb allows huge batch sizes without lossing accuracy via layer-wise adaptive update bounded by the inverse of its Lipschiz constant. Adafactor is a first-order Adam variant using Non-negative Matrix Factorization(NMF) to reduce memory footprint. CAME improves by introducting a confidence matrix to correct NMF. GaLore further reduces memory by projecting gradients into a low-rank space and 8-bit block-wise quantization. Lamb allows huge batch sizes without lossing accuracy via layer-wise adaptive update bounded by the inverse of its Lipschiz constant.
## API Reference
{{ autodoc:colossalai.nn.optimizer.distributed_adafactor.DistributedAdaFactor }}
{{ autodoc:colossalai.nn.optimizer.distributed_lamb.DistributedLamb }}
{{ autodoc:colossalai.nn.optimizer.distributed_galore.DistGaloreAwamW }}
{{ autodoc:colossalai.nn.optimizer.distributed_came.DistributedCAME }}
## Hands-On Practice ## Hands-On Practice
We now demonstrate how to use Distributed Adafactor with booster API combining Tensor Parallel and ZeRO 2 with 4 GPUs. **Note that even if you're not aware of distributed optimizers, the plugins automatically casts yours to the distributed version for convenience.** We now demonstrate how to use Distributed Adafactor with booster API combining Tensor Parallel and ZeRO 2 with 4 GPUs. **Note that even if you're not aware of distributed optimizers, the plugins automatically casts yours to the distributed version for convenience.**
@ -140,3 +134,10 @@ optim = DistGaloreAwamW(
</table> </table>
<!-- doc-test-command: colossalai run --nproc_per_node 4 distributed_optimizers.py --> <!-- doc-test-command: colossalai run --nproc_per_node 4 distributed_optimizers.py -->
## API Reference
{{ autodoc:colossalai.nn.optimizer.distributed_adafactor.DistributedAdaFactor }}
{{ autodoc:colossalai.nn.optimizer.distributed_lamb.DistributedLamb }}
{{ autodoc:colossalai.nn.optimizer.distributed_galore.DistGaloreAwamW }}
{{ autodoc:colossalai.nn.optimizer.distributed_came.DistributedCAME }}

13
docs/source/zh-Hans/features/distributed_optimizers.md
View File

@ -13,12 +13,6 @@ Author: Wenxuan Tan, Junwen Duan, Renjie Mao
## 优化器 ## 优化器
Adafactor 是一种首次采用非负矩阵分解（NMF）的 Adam 变体，用于减少内存占用。CAME 通过引入一个置信度矩阵来改进 NMF 的效果。GaLore 通过将梯度投影到低秩空间，并使用 8 位块状量化进一步减少内存占用。Lamb 允许使用巨大的批量大小而不失准确性，通过按其 Lipschitz 常数的倒数界定的逐层自适应更新实现 Adafactor 是一种首次采用非负矩阵分解（NMF）的 Adam 变体，用于减少内存占用。CAME 通过引入一个置信度矩阵来改进 NMF 的效果。GaLore 通过将梯度投影到低秩空间，并使用 8 位块状量化进一步减少内存占用。Lamb 允许使用巨大的批量大小而不失准确性，通过按其 Lipschitz 常数的倒数界定的逐层自适应更新实现
## API 参考
{{ autodoc:colossalai.nn.optimizer.distributed_adafactor.DistributedAdaFactor }}
{{ autodoc:colossalai.nn.optimizer.distributed_lamb.DistributedLamb }}
{{ autodoc:colossalai.nn.optimizer.distributed_galore.DistGaloreAwamW }}
{{ autodoc:colossalai.nn.optimizer.distributed_came.DistributedCAME }}
## 使用 ## 使用
现在我们展示如何使用分布式 Adafactor 与 booster API 结合 Tensor Parallel 和 ZeRO 2。即使您不使用distributed optimizer，plugin 也会自动将optimizer转换为分布式版本以方便使用。 现在我们展示如何使用分布式 Adafactor 与 booster API 结合 Tensor Parallel 和 ZeRO 2。即使您不使用distributed optimizer，plugin 也会自动将optimizer转换为分布式版本以方便使用。
@ -137,3 +131,10 @@ optim = DistGaloreAwamW(
</table> </table>
<!-- doc-test-command: colossalai run --nproc_per_node 4 distributed_optimizers.py --> <!-- doc-test-command: colossalai run --nproc_per_node 4 distributed_optimizers.py -->
## API 参考
{{ autodoc:colossalai.nn.optimizer.distributed_adafactor.DistributedAdaFactor }}
{{ autodoc:colossalai.nn.optimizer.distributed_lamb.DistributedLamb }}
{{ autodoc:colossalai.nn.optimizer.distributed_galore.DistGaloreAwamW }}
{{ autodoc:colossalai.nn.optimizer.distributed_came.DistributedCAME }}