Discussions: hpcaitech/ColossalAI

Discussions

• 1 You must be logged in to vote

  🙏

  我是否一定需要保证不同机器之间可以免密登陆才能使用多机训练

  yhcc asked Dec 20, 2022 in Q&A · Unanswered

  

  

  

  

  14
• 2 You must be logged in to vote

  🙏

  为什么用你们的框架训练出的stable diffusion ckpt 推理结果异常为噪声图

  yufengyao-lingoace asked Jan 12, 2023 in Q&A · Unanswered

  

  

  0
• 1 You must be logged in to vote

  🙏

  why can't I get a right output of stable diffusion follow your official document?

  yufengyao-lingoace asked Jan 11, 2023 in Q&A · Unanswered

  

  

  0
• 1 You must be logged in to vote

  💬

  colossal ai of stable diffusion outputs a picture with random noise

  yufengyao-lingoace started Jan 11, 2023 in General

  

  

  0
• 1 You must be logged in to vote

  💬

  单机finetune训练stable diffusion 需要的最低配置（内存和显存）是多少？

  yufengyao-lingoace started Dec 30, 2022 in General

  

  

  2
• 1 You must be logged in to vote

  🙏

  为什么32g内存16g显存batchsize=1跑不起来(stable diffusion )

  yufengyao-lingoace asked Jan 5, 2023 in Q&A · Unanswered

  

  

  5
• 1 You must be logged in to vote

  🙏

  Rotary Embeddings in Encoder

  conceptofmind asked Jan 6, 2023 in Q&A · Unanswered

  

  

  0
• 1 You must be logged in to vote

  🙏

  是否有更高效的大模型训练方法

  yhcc asked Jan 5, 2023 in Q&A · Answered

  

  

  

  

  38
• 2 You must be logged in to vote

  🙏

  pipeline并行支持动态修改batch_size和num_microbatches吗？

  yhcc asked Jan 4, 2023 in Q&A · Unanswered

  

  

  4
• 1 You must be logged in to vote

  🙏

  如何进行zero优化+TP+PP

  yhcc asked Dec 22, 2022 in Q&A · Unanswered

  

  

  

  

  6
• 2 You must be logged in to vote

  🙏

  Failed to build from source (pytorch 1.3.1 + CUDA 11.6)

  eric8607242 asked Jan 4, 2023 in Q&A · Answered

  

  

  1
• 1 You must be logged in to vote

  📣

  Version v0.2.0 Release Today!

  binmakeswell announced Jan 3, 2023 in Announcements

  

  

  0
• 1 You must be logged in to vote

  🙏

  使用colossalai原生的fp16是否会导致clip_grad_norm无效？

  yhcc asked Jan 1, 2023 in Q&A · Answered

  

  

  2
• 1 You must be logged in to vote

  🙏

  Can I use LowLevelZeroOptimizer in Engine so that I can use zero-1 and pipeline?

  yhcc asked Jan 2, 2023 in Q&A · Unanswered

  

  

  1
• 1 You must be logged in to vote

  🙏

  VocabParallelClassifier1D中设置是否有问题？

  yhcc asked Jan 2, 2023 in Q&A · Unanswered

  

  

  1
• 1 You must be logged in to vote

  🙏

  使用colossalai原生的fp16会在clip_grad_norm_fp32函数中卡死

  yhcc asked Jan 1, 2023 in Q&A · Unanswered

  

  

  1
• 1 You must be logged in to vote

  🙏

  How to save checkpoint?

  yhcc asked Dec 29, 2022 in Q&A · Unanswered

  

  

  0
• 1 You must be logged in to vote

  💬

  Support for Swin + AlBert

  nostalgicimp started Oct 18, 2022 in General

  

  

  

  

  3
• 1 You must be logged in to vote

  🙏

  Can't improve the scaler of batch size with ZeRO technique

  Lyn-Lucy asked Nov 10, 2022 in Q&A · Unanswered

  

  

  1
• 1 You must be logged in to vote

  🙏

  How to gather results from different process

  mrbeann asked Nov 30, 2022 in Q&A · Unanswered

  

  

  12
• 1 You must be logged in to vote

  🙏

  train_gpt_demo.py例子中的一些疑问

  yhcc asked Dec 22, 2022 in Q&A · Answered

  

  

  2
• 1 You must be logged in to vote

  🙏

  How does ZeRO DDP work with Tensor Parallelism?

  heya5 asked Dec 16, 2022 in Q&A · Unanswered

  

  

  1
• 1 You must be logged in to vote

  💬

  Support for CUDA 11.2?

  TearGosling started Dec 12, 2022 in General

  

  

  0
• 6 You must be logged in to vote

  🙏

  What's the RAM requirement of the Stable Diffusion example?

  CCRcmcpe asked Nov 10, 2022 in Q&A · Unanswered

  

  

  

  

  2
• 1 You must be logged in to vote

  🙏

  Where are the examples for https://arxiv.org/pdf/2110.14883.pdf？

  tongping asked Dec 7, 2022 in Q&A · Unanswered

  

  

  2