GGitHubAdd DeepSpeed finetune demo
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
DeepSpeedExamples major restructure (#245) | 3 年前 | |
DeepSpeedExamples major restructure (#245) | 3 年前 | |
Update domino example (#976) * remove files Signed-off-by: Hongwei Chen <hongweichen@microsoft.com> * Update domino example Signed-off-by: Hongwei Chen <hongweichen@microsoft.com> * apply review suggestions Signed-off-by: Hongwei Chen <hongweichen@microsoft.com> --------- Signed-off-by: Hongwei Chen <hongweichen@microsoft.com> | 1 年前 | |
Superoffload examples (#990) * feat: add examples for superoffload * fix: typo * fix: remove hardcoded GPU bind * feat: add requirement for superoffload | 11 个月前 | |
Fix README for LLaMA-2 fine-tuning with ZenFlow. (#987) Signed-off-by: Tingfeng Lan <erc8gx@virginia.edu> | 1 年前 | |
Fix errors of AttributeError: 'str' object has no attribute 'stdout' (#826) * Update train_bert.py Fix errors of stdout * Update train_bert.py fix errors of stdout --------- Co-authored-by: Michael Wyatt <mrwyattii@gmail.com> | 2 年前 | |
Update requirements for opencv-python CVE (#925) * Update requirements for opencv-python CVE * Dont' update numpy for now | 1 年前 | |
Update references to deepspeedai GH org (#955) Signed-off-by: Logan Adams <loadams@microsoft.com> | 1 年前 | |
Add comment explaining nesting `torch.autocast` (#1000) * Add comment explaining outer torch.autocast in bf16_master_weight example The outer autocast covers loss_fn which runs outside engine.forward(). The nested autocast on the model forward is harmless. Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * Update training/bf16_master_weight/train.py Co-authored-by: Stas Bekman <stas00@users.noreply.github.com> --------- Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> Co-authored-by: Stas Bekman <stas00@users.noreply.github.com> | 6 个月前 | |
Update references to deepspeedai GH org (#955) Signed-off-by: Logan Adams <loadams@microsoft.com> | 1 年前 | |
Update references to torchvision (#949) | 1 年前 | |
Variable batch size and LR example for DeepSpeed PR #7104 (#963) * moved example from DeepSpeed PR #7104 to this repo * Update training/data_efficiency/variable_batch_size_and_lr/README.md Co-authored-by: Olatunji Ruwase <olruwase@microsoft.com> * Update training/data_efficiency/variable_batch_size_and_lr/README.md Co-authored-by: Olatunji Ruwase <olruwase@microsoft.com> * replaced T by S for sequence length * replaced T by S for sequence length * replaced T by S for sequence length * more detailed explanation * --pipeline-num-stages is now a comd line argument * cleaner syntax * Update training/data_efficiency/variable_batch_size_and_lr/README.md --------- Co-authored-by: Olatunji Ruwase <olruwase@microsoft.com> | 1 年前 | |
Add DeepSpeed finetune demo DeepSpeed finetune demo aims at to be a one stop demo for DeepSpeed training features. By using the same training script with different deepspeed configurations, we are able to compare different training parallelism side-by-side with convergence, throughput, and downstream tasks. | 3 个月前 | |
Update references to torchvision (#949) | 1 年前 | |
fix: the json format of the training imagenet configuration file (#954) | 1 年前 | |
Update references to deepspeedai GH org (#955) Signed-off-by: Logan Adams <loadams@microsoft.com> | 1 年前 | |
Example and benchmark of APIs to offload states (#942) * add benchmarking for offloading states * fix api names | 1 年前 | |
Update references to torchvision (#949) | 1 年前 | |
Adding LoRA-Distillation SD training example (#788) Co-authored-by: Xiaoxia (Shirley) Wu <94406484+xiaoxiawu-microsoft@users.noreply.github.com> | 2 年前 | |
Add examples of AutoTP (#998) * add custom autotp example Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * update key name Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * update tp examples Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * update tp examples Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * update example Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * update READMEs Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * replace deprecated openai package Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * fix patterns Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> * update train config Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> --------- Signed-off-by: Masahiro Tanaka <mtanaka@anyscale.com> | 6 个月前 |