github
/
ColossalAI
mirror of https://github.com/hpcaitech/ColossalAI
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[nfc] fix typo in docs/ (#4972)

pull/5089/head
digger yu 2023-11-21 22:06:20 +08:00 committed by GitHub
parent dce05da535
commit fd3567e089
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
docs/source/en/basics/booster_plugins.md
View File

@ -26,7 +26,7 @@ Generally only one plugin is used to train a model. Our recommended use case for
- [Torch DDP Plugin](#torch-ddp-plugin): It is suitable for models with less than 2 billion parameters (e.g. Bert-3m, GPT2-1.5b).
- [Torch FSDP Plugin](#torch-fsdp-plugin) / [Low Level Zero Plugin](#low-level-zero-plugin): It is suitable for models with less than 10 billion parameters (e.g. GPTJ-6b, MegatronLM-8b).
- [Gemini Plugin](#gemini-plugin): It is suitable for models with more than 10 billion parameters (e.g. TuringNLG-17b) and is ideal for scenarios with **high cross-node bandwidth and medium to small-scale clusters (below a thousand cards)** (e.g. Llama2-70b).
- [Hybrid Pararllel Plugin](#hybrid-parallel-plugin): It is suitable for models with more than 60 billion parameters, or special models such as those with exceptionally long sequences, very large vocabularies, and is best suited for scenarios with **low cross-node bandwidth and large-scale clusters (a thousand cards or more)** (e.g. GPT3-175b, Bloom-176b).
- [Hybrid Parallel Plugin](#hybrid-parallel-plugin): It is suitable for models with more than 60 billion parameters, or special models such as those with exceptionally long sequences, very large vocabularies, and is best suited for scenarios with **low cross-node bandwidth and large-scale clusters (a thousand cards or more)** (e.g. GPT3-175b, Bloom-176b).
## Plugins

4
docs/source/zh-Hans/basics/booster_plugins.md
View File

@ -16,7 +16,7 @@
- [Torch FSDP 插件](#torch-fsdp-插件): 它包装了 `torch.distributed.fsdp.FullyShardedDataParallel` 并且可用于使用 Zero-dp 训练模型。
- [Low Level Zero 插件](#low-level-zero-插件): 它包装了 `colossalai.zero.low_level.LowLevelZeroOptimizer`，可用于使用 Zero-dp 训练模型。它仅支持 Zero 阶段1和阶段2。
- [Gemini 插件](#gemini-插件): 它包装了 [Gemini](../features/zero_with_chunk.md)，Gemini 实现了基于Chunk内存管理和异构内存管理的 Zero-3。
- [Hybrid Pararllel 插件](#hybrid-parallel-插件): 它为Shardformer，流水线管理器，混合精度运算，TorchDDP以及Zero-1/Zero-2功能提供了一个统一且简洁的接口。使用该插件可以简单高效地实现transformer模型在张量并行，流水线并行以及数据并行（DDP, Zero）间任意组合并行训练策略，同时支持多种训练速度和内存的优化工具。有关这些训练策略和优化工具的具体信息将在下一章中阐述。
- [Hybrid Parallel 插件](#hybrid-parallel-插件): 它为Shardformer，流水线管理器，混合精度运算，TorchDDP以及Zero-1/Zero-2功能提供了一个统一且简洁的接口。使用该插件可以简单高效地实现transformer模型在张量并行，流水线并行以及数据并行（DDP, Zero）间任意组合并行训练策略，同时支持多种训练速度和内存的优化工具。有关这些训练策略和优化工具的具体信息将在下一章中阐述。
更多插件即将推出。
@ -24,7 +24,7 @@
- [Torch DDP 插件](#torch-ddp-插件): 适用于参数少于 20 亿的模型（例如 Bert-3m、GPT2-1.5b）。
- [Torch FSDP 插件](#torch-fsdp-插件) / [Low Level Zero 插件](#low-level-zero-插件): 适用于参数少于 100 亿的模型（例如 GPTJ-6b、MegatronLM-8b）。
- [Gemini 插件](#gemini-插件): 适合参数超过 100 亿的模型（例如 TuringNLG-17b），且**跨节点带宽高、中小规模集群（千卡以下）**的场景（例如 Llama2-70b）。
- [Hybrid Pararllel 插件](#hybrid-parallel-插件): 适合参数超过 600 亿的模型、超长序列、超大词表等特殊模型，且**跨节点带宽低、大规模集群（千卡以上）**的场景（例如 GPT3-175b、Bloom-176b）。
- [Hybrid Parallel 插件](#hybrid-parallel-插件): 适合参数超过 600 亿的模型、超长序列、超大词表等特殊模型，且**跨节点带宽低、大规模集群（千卡以上）**的场景（例如 GPT3-175b、Bloom-176b）。
## 插件