Issues

Issues

• Compute_rewards in PPO:rewards[j, start:ends[j]][-1] += reward_clip[j] is wrong

  deepspeedai/DeepSpeedExamples#996 · opened on Dec 22, 2025

• Step1 failed when I run "bash training_scripts/opt/single_gpu/run_1.3b.sh"

  deepspeedai/DeepSpeedExamples#995 · opened on Dec 16, 2025

• One example, multiple config files

  deepspeedai/DeepSpeedExamples#989 · opened on Sep 8, 2025

• moe example 404

  deepspeedai/DeepSpeedExamples#984 · opened on Jul 25, 2025

• Possible to include an example of DeepNVMe + state dict

  deepspeedai/DeepSpeedExamples#979 · opened on Jun 19, 2025
  6

• Apply Zero-3 and LoRA appears empty lora weight [0]

  deepspeedai/DeepSpeedExamples#969 · opened on Apr 16, 2025
  1

• DeepSpeed-FastGen support ascend npu?

  deepspeedai/DeepSpeedExamples#960 · opened on Mar 4, 2025
  3

• Why Does vf_loss Take the Maximum Value, Rendering Clamp Meaningless?

  deepspeedai/DeepSpeedExamples#956 · opened on Feb 7, 2025
  2

• Is there any example about DeepSpeed Zero with Ulysses/Ulysses-offload

  deepspeedai/DeepSpeedExamples#951 · opened on Jan 10, 2025

• Domino + PP

  deepspeedai/DeepSpeedExamples#950 · opened on Jan 9, 2025

• Assertion srcIndex < srcSelectDimSize failed

  deepspeedai/DeepSpeedExamples#946 · opened on Dec 18, 2024
  1

• Question to attention computation

  deepspeedai/DeepSpeedExamples#944 · opened on Dec 15, 2024

• KV_cache offload

  deepspeedai/DeepSpeedExamples#943 · opened on Dec 15, 2024
  3

• How can I change the master_port when using deepspeed for multi-GPU on single node, i.e. localhost

  deepspeedai/DeepSpeedExamples#936 · opened on Nov 20, 2024
  4

• Does DeepSpeed's Pipeline-Parallelism optimizer supports skip connections?

  deepspeedai/DeepSpeedExamples#932 · opened on Oct 17, 2024

• How to calculate training efficiency ,i.e tokens/sec of step 1 fine tuning of llama2 model ?

  deepspeedai/DeepSpeedExamples#923 · opened on Aug 29, 2024

• Actor loss nan and Resizing model embedding

  deepspeedai/DeepSpeedExamples#922 · opened on Aug 29, 2024
  1

• step2 without any response for a long time

  deepspeedai/DeepSpeedExamples#915 · opened on Aug 15, 2024

• Different zero stage the training memory compute

  deepspeedai/DeepSpeedExamples#912 · opened on Jul 17, 2024

• an error with gradient checkpointing in DeepspeedChat step 3

  deepspeedai/DeepSpeedExamples#908 · opened on Jun 28, 2024

• 单机多卡进行RLHF在第三步中使用Qwen模型作Actor Model报错

  deepspeedai/DeepSpeedExamples#907 · opened on Jun 25, 2024
  2

• DeepSpeed-Chat step-1 hanging for a long time

  deepspeedai/DeepSpeedExamples#906 · opened on Jun 19, 2024

• CPU OOM when inferencing Llama3-70B-Chinese-Chat

  deepspeedai/DeepSpeedExamples#904 · opened on May 20, 2024

• cannot pickle 'Stream' object

  deepspeedai/DeepSpeedExamples#903 · opened on May 17, 2024

• can not run the test-gpt.sh because of assertionError

  deepspeedai/DeepSpeedExamples#902 · opened on May 16, 2024