Skip to content

Issues: hpcaitech/ColossalAI

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

418 Open 1,287 Closed
418 Open 1,287 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

[BUG]: 用64卡或者80卡训练 loss 总是nan bug Something isn't working
#6239 opened Mar 6, 2025 by xujin1184104394
2 tasks done
5
[DOC]: 可以提供使用ColossalAI 训练自定义的模型 documentation Improvements or additions to documentation
#6238 opened Mar 6, 2025 by lianNice
1
[BUG]: raise ValueError("No parameters found in moe_dp_group, please consider using HybridParallelPlugin instead") ValueError: No parameters found in moe_dp_group, please consider using HybridParallelPlugin instead bug Something isn't working
#6237 opened Mar 4, 2025 by klompn
2 tasks done
3
[DOTry to use lora_finetune.py directly and get an error /bin/bash: line 1: export: `=/usr/bin/supervisord': not a valid identifier documentation Improvements or additions to documentation
#6235 opened Mar 2, 2025 by tanghl01
[BUG]: cannot import name 'CpuAdamArmExtension' from 'colossalai.kernel.extensions' (unknown location) bug Something isn't working
#6233 opened Feb 28, 2025 by Fence
2 tasks done
DeepSeekV3 lora fine tune, allocate memory too big !
#6224 opened Feb 26, 2025 by wotulong
1
[BUG]: Lora load error bug Something isn't working
#6221 opened Feb 25, 2025 by 447428054
2 tasks done
[BUG]: EP16 negative split bug Something isn't working
#6220 opened Feb 25, 2025 by 447428054
2 tasks done
【Question】What is the minimum number of GPUs required to train deepseek 671B with GRPO? How about using LoRA?
#6219 opened Feb 25, 2025 by LiuShixing
[BUG]: /bin/bash: line 0: export: `NPU-VISIBLE-DEVICES=0,1,2,3,4,5,6,7': not a valid identifier bug Something isn't working
#6217 opened Feb 24, 2025 by Gera001
2 tasks done
2
Respecting regulations and stabilizing the ecosystem by activists bug Something isn't working
#6216 opened Feb 24, 2025 by MASIHMIRSALI
2 tasks done
1
[BUG]: Precision overflow occurs when moe forward is performed bug Something isn't working
#6212 opened Feb 21, 2025 by zh2333
2 tasks done
2
[BUG]: failed to install coati in npu docker environment bug Something isn't working
#6209 opened Feb 20, 2025 by wangyuan249
2 tasks done
2
[BUG]: 该如何安装colossal到NPU上，看项目有相关描述，但没找到相关教程 bug Something isn't working
#6205 opened Feb 20, 2025 by obj12
2 tasks done
4
[BUG]: bug Something isn't working
#6202 opened Feb 19, 2025 by Hongyuan-Liu
2 tasks done
1
【Question】Question about initial finetune loss
#6201 opened Feb 19, 2025 by SolenoidWGT
2
[DOC]: Update the documentation of ShardConfig for 1D, 2D, 2.5D, 3D tensor parallelism documentation Improvements or additions to documentation
#6197 opened Feb 18, 2025 by giriprasad51
it free?
#6193 opened Feb 15, 2025 by werruww
[FEATURE]: Expert Parallel for qwen/deepseek enhancement New feature or request
#6180 opened Jan 12, 2025 by Guodanding
4
[BUG]: RuntimeError: mat1 and mat2 must have the same dtype, but got Float and BFloat16 bug Something isn't working
#6169 opened Dec 25, 2024 by balcklive
1 task done
Hybrid Parallel Plugin下TP显存比同配置下deepspeed要高？？？
#6161 opened Dec 17, 2024 by duomicoding
5
[BUG]: Gemini saved an additional portion of the weights while using tie_word_embeddings=True bug Something isn't working
#6160 opened Dec 13, 2024 by ericxsun
1 task done
update shardformer for transformers=4.46
#6157 opened Dec 10, 2024 by duomicoding
[FEATURE]: Lora/QLora in GeminiPlugin and TorchFSDP enhancement New feature or request
#6138 opened Nov 16, 2024 by ericxsun
1
[FEATURE]: support google/gemma-2-2b for Tensor Parallelism enhancement New feature or request
#6120 opened Nov 9, 2024 by jing-4369
2
ProTip! Type g i on any issue or pull request to go back to the issue listing page.