deepspeedai/DeepSpeedExamples

View on GitHub

Example models using DeepSpeed

Stars: 6.8k
Forks: 1.1k
Open beginner issues: 0
Indexed issues: 300
Avg merge: 2d 16h
Merged PRs (30d): 1

Dominant language: Python
License: Apache-2.0
Last GitHub push: Aug 26, 2026
Latest indexed: Sep 20, 2026
Contributing guide: No contributing guide
Code of conduct: Code of conduct
Beginner labels: No beginner labels indexed

Beginner friendly Everything

Open All

300 open issues indexed

[ERROR]In Step3，load reward Model failed which trainged with zero-stage 3 Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#429 · 2 reactions · 1 assignee · Opened Apr 25, 2023
Step 3 1.3b Running process stuck Open

deespeed chat new-config system

deepspeedai/DeepSpeedExamples#428 · 4 comments · 1 assignee · Opened Apr 25, 2023
How can I train step3 in DeepSpeed-Chat by pipeline parallelism? Open

deespeed chat question

deepspeedai/DeepSpeedExamples#427 · 1 comment · 1 reaction · 1 assignee · Opened Apr 25, 2023
Bug: incorrect metrics evaluating for step two Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#425 · 3 comments · 1 assignee · Opened Apr 25, 2023
step3 failed actor opt_1.3b critic opt_350m Exception: Current loss scale already at minimum - cannot decrease scale anymore. Exiting run Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#419 · 1 comment · 1 assignee · Opened Apr 25, 2023
Cannot load the previous model weights when using ZeRO 3 optimizer in DeepSpeed Chat Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#417 · 4 comments · 2 assignees · Opened Apr 25, 2023
Error after changing the model from opt to gpt Open

deespeed chat hybrid engine new-config

deepspeedai/DeepSpeedExamples#403 · 2 comments · 1 assignee · Opened Apr 23, 2023
SFT training ,single gpu (V100 32G), how to adjust my parameters to avoid OOM, thx Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#389 · 3 comments · 1 assignee · Opened Apr 21, 2023
Default configuration running with V100-32G causes OOM Open

deespeed chat system

deepspeedai/DeepSpeedExamples#387 · 7 comments · 2 assignees · Opened Apr 21, 2023
Step 3: RuntimeError: CUDA error: misaligned address Open

deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#385 · 7 comments · 1 assignee · Opened Apr 21, 2023
Currently, the docker image is suitable for Ubuntu. Could you provide a method for CentOS? Open

deepspeedai/DeepSpeedExamples#382 · 4 comments · 1 assignee · Opened Apr 21, 2023
Run multi-node training failed, how to train without hostfile Open

deespeed chat system

deepspeedai/DeepSpeedExamples#381 · 5 comments · 1 assignee · Opened Apr 21, 2023
AttributeError: 'DeepSpeedHybridEngine' object has no attribute 'mp_group' in step 3. Open

deespeed chat hybrid engine

Difficulty 4/5 3-5 days Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#379 · 9 comments · Opened Apr 21, 2023
【BUG】occur error：AttributerError：'DeepSpeedHybridEngine' object has no attribute 'mp_group' whiling run llama7b for step3/rlhf/ppo Open

deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#375 · 3 comments · 1 assignee · Opened Apr 20, 2023
Missing key(s) in state_dict for bias in attention blocks Open

bug deespeed chat

deepspeedai/DeepSpeedExamples#374 · 3 comments · 1 reaction · 1 assignee · Opened Apr 20, 2023
When running Stage-3 scripts with enable_hybrid_engine encountered errors Open

deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#373 · 9 comments · 4 reactions · 2 assignees · Opened Apr 20, 2023
Exchange group; 交流群 Open

deespeed chat question

Difficulty 5/5 Over a week Newbie friendliness 15/100

deepspeedai/DeepSpeedExamples#367 · 8 comments · 6 reactions · Opened Apr 20, 2023
how to save model, i cant load saved llama7b model Open

deespeed chat llama

deepspeedai/DeepSpeedExamples#365 · 1 comment · 1 assignee · Opened Apr 20, 2023
map[i] = val_or_map.get(i, Std.NONE) AttributeError: 'NoneType' object has no attribute 'get' Open

bug deespeed chat new-config

Difficulty 3/5 1-2 days Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#359 · 1 comment · Opened Apr 19, 2023
use bloom-350m to train reward model in step2 Open

deespeed chat new-config

Difficulty 4/5 3-5 days Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#356 · 7 comments · Opened Apr 19, 2023
it took much time to initial:Initializing TorchBackend in DeepSpeed with backend nccl. Could you please help me to eliminate the consumed time.thx Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

deepspeedai/DeepSpeedExamples#355 · 3 comments · Opened Apr 19, 2023
llama7b in v100*2（32gb） OOM Open

Difficulty 4/5 3-5 days Newbie friendliness 15/100

deepspeedai/DeepSpeedExamples#354 · Opened Apr 19, 2023
About release date for Llama system support Open

deespeed chat llama

Difficulty 5/5 Over a week Newbie friendliness 12/100

deepspeedai/DeepSpeedExamples#350 · 2 comments · 9 reactions · Opened Apr 19, 2023
Using LLaMA in reward model training Open

deespeed chat llama

Difficulty 4/5 3-5 days Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#349 · 6 comments · Opened Apr 19, 2023
Job hang at model forward for rank 0 after saving immediate ckpt, multi nodes, ZeRO3 Open

deespeed chat llama

deepspeedai/DeepSpeedExamples#344 · 3 comments · 2 assignees · Opened Apr 18, 2023
Could this tool apply for encoder-decoder model, like Flan-T5? Open

deespeed chat new-config

Difficulty 5/5 Over a week Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#340 · 4 comments · 1 reaction · Opened Apr 18, 2023
多机分布式训练，加载模型，报a leaf Variable that requires grad is being used in an in-place operation错误 Open

bug deespeed chat

Difficulty 4/5 3-5 days Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#339 · 2 comments · Opened Apr 18, 2023
Error when using BLOOMZ for reward model training Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#338 · 16 comments · 1 assignee · Opened Apr 18, 2023
DeepSpeed-Chat: prefetch of layers during reward model forward pass leads to error during sample generation Open

bug deespeed chat

deepspeedai/DeepSpeedExamples#337 · 27 comments · 1 assignee · Opened Apr 18, 2023
single gpu 6.7b lora CUDA OOM with A6000 48G Open

deespeed chat question

deepspeedai/DeepSpeedExamples#330 · 4 comments · 1 assignee · Opened Apr 17, 2023
Step1 training failed Open

bug deespeed chat system

deepspeedai/DeepSpeedExamples#328 · 1 comment · 1 assignee · Opened Apr 17, 2023
Running multinode training and received unclear error for stage 2 training Open

bug deespeed chat system

deepspeedai/DeepSpeedExamples#327 · 4 comments · 1 assignee · Opened Apr 17, 2023
[BUG]Step1 RuntimeError: CUDA error: CUBLAS_STATUS_ALLOC_FAILED when calling `cublasCreate(handle)` Open

bug deespeed chat system

deepspeedai/DeepSpeedExamples#323 · 1 comment · 1 assignee · Opened Apr 17, 2023
Step2 training get a negative score and accuray is below 60% Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#322 · 1 comment · 1 assignee · Opened Apr 17, 2023
Step2: memory allocation of 2097152 bytes failed Open

bug deespeed chat

deepspeedai/DeepSpeedExamples#321 · 4 comments · 1 assignee · Opened Apr 17, 2023
In instructGPT, during the RM training process, different <prompt, response> pairs of a prompt are put together to calculate the loss. Is this also implemented in DeepSpeed-chat? Open

deespeed chat question

deepspeedai/DeepSpeedExamples#320 · 1 assignee · Opened Apr 17, 2023
Model performance suprisingly bad Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#318 · 9 comments · 2 reactions · 1 assignee · Opened Apr 16, 2023
run deepspeed_chat example code error Open

bug deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#313 · 4 comments · 1 assignee · Opened Apr 15, 2023
DeepSpeed op builder Open

Difficulty 5/5 Over a week Newbie friendliness 10/100

deepspeedai/DeepSpeedExamples#312 · 4 comments · Opened Apr 15, 2023
when I am running RLHF script, I encountered a error Open

bug deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#311 · 3 comments · 1 assignee · Opened Apr 15, 2023
Single node multi card training failed Open

bug deespeed chat system

deepspeedai/DeepSpeedExamples#310 · 1 comment · 1 assignee · Opened Apr 15, 2023
My model Performs Badly...Is GPU memory to small? Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#307 · 12 comments · 2 reactions · 1 assignee · Opened Apr 15, 2023
Does the model in this framework have to be trained to have conversations? Open

deespeed chat modeling question

Difficulty 5/5 Over a week Newbie friendliness 15/100

deepspeedai/DeepSpeedExamples#306 · 1 comment · Opened Apr 15, 2023
If I use a self-improved transformer architecture, can it support? Open

deespeed chat new-config question

Difficulty 5/5 Over a week Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#304 · Opened Apr 14, 2023
Feature Request: Support for Encoder Deocder models like T5 Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

deepspeedai/DeepSpeedExamples#265 · 1 reaction · Opened Apr 12, 2023
ValueError: optimizer got an empty parameter list Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#259 · Opened Apr 6, 2023
The model size does not change Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

deepspeedai/DeepSpeedExamples#239 · 3 comments · Opened Mar 1, 2023
How can I find bert weights used for ZeroQuant? Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

deepspeedai/DeepSpeedExamples#236 · Opened Feb 8, 2023
AttributeError: '_MultiProcessingDataLoaderIter' object has no attribute 'next' Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

deepspeedai/DeepSpeedExamples#222 · 3 comments · Opened Nov 17, 2022
autotuning, error Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

deepspeedai/DeepSpeedExamples#216 · 1 comment · Opened Nov 10, 2022

deepspeedai/DeepSpeedExamples

Get new issues in your inbox