TechQA.

Question

I met problem when installing DeepSpeed from source

score 64 · Answer 1 · 2024-03-15T09:10:38.567000

0

Answer

64

Views

I met problem when installing DeepSpeed from source

64 views Asked by user23623119 At 15 March 2024 at 09:10

score 144 · Answer 2 · 2024-03-13T11:25:22.660000

Using uv to install packages in the bitnami/deepspeed:0.14.0 Docker image fails with 'uv: command not found'

144 views Asked by BioGeek At 13 March 2024 at 11:25

score 164 · Answer 3 · 2024-02-20T11:28:33.470000

Problems when profiling LLM-training using "huggingface/accelerate" to Night system

164 views Asked by 상현박 At 20 February 2024 at 11:28

score 40 · Answer 4 · 2024-02-15T22:10:53.800000

deespeed getting output shape wrong on stages>1

40 views Asked by Romeo Kienzler At 15 February 2024 at 22:10

score 651 · Answer 5 · 2023-12-27T05:12:41.763000

pip install deepspeed ERROR: error: subprocess-exited-with-error/error: metadata-generation-failed

651 views Asked by TTTyz At 27 December 2023 at 05:12

score 202 · Answer 6 · 2023-12-19T12:31:00.713000

Exits with return code = -9 when pretrain llama2

202 views Asked by Jim At 19 December 2023 at 12:31

score 787 · Answer 7 · 2023-11-22T18:38:47.453000

LLava: deepspeed can not detect editable installed python package/module

787 views Asked by Mohbat Tharani At 22 November 2023 at 18:38

score 179 · Answer 8 · 2023-11-16T02:04:50.857000

why accelerate need Multiply accelerator.num_processes

179 views Asked by TuoMin At 16 November 2023 at 02:04

score 2266 · Answer 9 · 2023-11-08T05:08:38.357000

DeepSpeed multi-GPU finetuning does not work

2.2k views Asked by kopilot100 At 08 November 2023 at 05:08

score 342 · Answer 10 · 2023-11-01T02:42:52.023000

Deepspeed not offloading to CPU

342 views Asked by paragon00 At 01 November 2023 at 02:42

score 181 · Answer 11 · 2023-10-11T14:43:25.727000

'MPTConfig' object has no attribute 'hidden_size'

181 views Asked by sanminchui At 11 October 2023 at 14:43

score 452 · Answer 12 · 2023-08-15T12:59:28.417000

You are using ZeRO-Offload with a client provided optimizer (<class 'torch.optim.adamw.AdamW'>) which in most cases will yield poor performance

452 views Asked by Zachary Nagler At 15 August 2023 at 12:59

score 1177 · Answer 13 · 2023-08-13T18:33:21.103000

Loading a HF Model in Multiple GPUs and Run Inferences in those GPUs (Not Training or Finetuning)

1.1k views Asked by NeuralAI At 13 August 2023 at 18:33

score 190 · Answer 14 · 2023-08-02T13:28:03.453000

Does Vertex AI Training for Distributed Training Across Multi-Nodes Work With HuggingFace Trainer + Deepspeed?

190 views Asked by esdy At 02 August 2023 at 13:28

score 294 · Answer 15 · 2023-07-30T06:32:11.373000

Deepspeed tensor parallel gets problem in tensor alignment when using tokenizer

294 views Asked by ddaa At 30 July 2023 at 06:32

score 111 · Answer 16 · 2023-07-29T14:10:13.297000

Why does the DeepSpeed `estimate_zero2_model_states_mem_needs_…` API report the same memory per CPU with different `offload_optimizer` option values?

score 234 · Answer 17 · 2023-07-28T04:51:25.970000

Training time for dolly-v2-12b on a custom dataset with an A10 gpu

234 views Asked by Sneha T S At 28 July 2023 at 04:51

score 116 · Answer 18 · 2023-07-24T07:39:43.377000

how to set max gpu memory use for each device when using deepspeed for distributed training?

116 views Asked by hjc At 24 July 2023 at 07:39

score 460 · Answer 19 · 2023-07-18T09:12:31.093000

How can I use decaying learning rate in DeepSpeed?

460 views Asked by AndyLinOuO At 18 July 2023 at 09:12

score 419 · Answer 20 · 2023-07-06T18:32:10.173000

How to add Deepspeed Activation Checkpointing to LLM for Fine-Tuning in PyTorch Lightning?

419 views Asked by Riley Hun At 06 July 2023 at 18:32

TechQA.

List Question

I met problem when installing DeepSpeed from source

Using uv to install packages in the bitnami/deepspeed:0.14.0 Docker image fails with 'uv: command not found'

Problems when profiling LLM-training using "huggingface/accelerate" to Night system

deespeed getting output shape wrong on stages>1

pip install deepspeed ERROR: error: subprocess-exited-with-error/error: metadata-generation-failed

Exits with return code = -9 when pretrain llama2

LLava: deepspeed can not detect editable installed python package/module

why accelerate need Multiply accelerator.num_processes

DeepSpeed multi-GPU finetuning does not work

Deepspeed not offloading to CPU

'MPTConfig' object has no attribute 'hidden_size'

You are using ZeRO-Offload with a client provided optimizer (<class 'torch.optim.adamw.AdamW'>) which in most cases will yield poor performance

Loading a HF Model in Multiple GPUs and Run Inferences in those GPUs (Not Training or Finetuning)

Does Vertex AI Training for Distributed Training Across Multi-Nodes Work With HuggingFace Trainer + Deepspeed?

Deepspeed tensor parallel gets problem in tensor alignment when using tokenizer

Why does the DeepSpeed `estimate_zero2_model_states_mem_needs_…` API report the same memory per CPU with different `offload_optimizer` option values?

Training time for dolly-v2-12b on a custom dataset with an A10 gpu

how to set max gpu memory use for each device when using deepspeed for distributed training?

How can I use decaying learning rate in DeepSpeed?

How to add Deepspeed Activation Checkpointing to LLM for Fine-Tuning in PyTorch Lightning?

Popular Questions

Trending Questions