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

step3_rlhf_finetuning and two tokenizers Open

deespeed chat modeling new-config question

deepspeedai/DeepSpeedExamples#577 · 3 comments · 1 reaction · 1 assignee · Opened Jun 6, 2023
step2 bug fix for loss = nan when using BLOOM(which is left padding style) Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#571 · 1 assignee · Opened Jun 2, 2023
messy response from model trained with opt-1.3b and Dahoas/rm-static Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#569 · 1 comment · 1 assignee · Opened Jun 2, 2023
Load model error in step3 Open

bug deespeed chat

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

deepspeedai/DeepSpeedExamples#560 · 8 comments · Opened May 31, 2023
7B bloom cuda oom run step3 with 80 T4(16G) when hybridEngine is on, but ok when hybridengine is off Open

deespeed chat hybrid engine

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

deepspeedai/DeepSpeedExamples#559 · Opened May 31, 2023
【problem discuss】Critic Loss can not decrease Open

deespeed chat modeling

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

deepspeedai/DeepSpeedExamples#556 · 17 comments · Opened May 30, 2023
what data should I use in step 3 Open

deespeed chat modeling question

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

deepspeedai/DeepSpeedExamples#555 · Opened May 29, 2023
【Need Help】What is [state, action, reward ] in NLP Scenario for PPO in deepspeed-chat Open

deespeed chat modeling

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

deepspeedai/DeepSpeedExamples#552 · 1 comment · Opened May 26, 2023
Model and data downloading Open

deespeed chat question

Difficulty 2/5 1-3 hours Newbie friendliness 45/100

deepspeedai/DeepSpeedExamples#550 · 2 comments · Opened May 26, 2023
how to apply DeepSpeedExamples with mpirun ? Open

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

deepspeedai/DeepSpeedExamples#548 · 1 comment · 1 reaction · Opened May 25, 2023
step3 answer is not correct Open

deespeed chat modeling

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

deepspeedai/DeepSpeedExamples#547 · 5 comments · Opened May 25, 2023
Recommended base docker image, torch/cuda version to use that is compatible with this code base? Open

deespeed chat system

Difficulty 3/5 1-2 days Newbie friendliness 30/100

deepspeedai/DeepSpeedExamples#546 · Opened May 23, 2023
When to support the llama model? Open

deespeed chat llama

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

deepspeedai/DeepSpeedExamples#540 · 1 comment · Opened May 21, 2023
The min_length setting force the model generate to max length, which produce repeated or nonsense result Open

deespeed chat modeling

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

deepspeedai/DeepSpeedExamples#539 · 1 comment · 1 reaction · Opened May 20, 2023
Hyper-param tuning for PPO Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#532 · 1 comment · 1 assignee · Opened May 17, 2023
When running step3, the error "CUDA error: misaligned address"? Open

deespeed chat

Difficulty 4/5 3-5 days Newbie friendliness 18/100

deepspeedai/DeepSpeedExamples#530 · 1 comment · 1 reaction · Opened May 17, 2023
Much more memory used in step 3 when using multi gpus compared to using single gpu Open

deespeed chat llama system

deepspeedai/DeepSpeedExamples#529 · 5 comments · 1 assignee · Opened May 16, 2023
Rewards in ppo seem to be recomputed many times Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#528 · 2 comments · 1 assignee · Opened May 16, 2023
[bug]AttributeError: 'DeepSpeedHybridEngine' object has no attribute 'mp_group' Open

bug deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#525 · 4 comments · 1 assignee · Opened May 15, 2023
OOM problem when fine-tune reward model with LLaMA in step 2 Open

deespeed chat llama

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

deepspeedai/DeepSpeedExamples#521 · 1 comment · Opened May 11, 2023
RLHF model return '{: {: {:' of every input Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#518 · 4 comments · 1 assignee · Opened May 11, 2023
In Step3, RuntimeError:RewardModel:size mismatch for rwtranrsformer.decoder.embed_tokens.weight Open

deespeed chat llama

deepspeedai/DeepSpeedExamples#512 · 6 comments · 1 assignee · Opened May 10, 2023
DeepSpeed-Chat cannot load models from local file? Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#511 · 4 comments · 1 assignee · Opened May 10, 2023
How to save the model after each epoch Open

deespeed chat question

Difficulty 4/5 3-5 days Newbie friendliness 20/100

deepspeedai/DeepSpeedExamples#510 · 1 comment · Opened May 10, 2023
Is there any Deepspeed Inference PTQ Example? Open

deespeed chat question

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

deepspeedai/DeepSpeedExamples#508 · Opened May 10, 2023
Eextension of the issus #479, chatbot.py cannot load the bloom model Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#505 · 3 comments · 1 assignee · Opened May 10, 2023
Might be a bug of hibrid engine : In Step3 wrong generation secquence when hibrid engine is enabled. Open

deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#503 · 7 comments · 1 reaction · 1 assignee · Opened May 9, 2023
CUDA error: CUBLAS_STATUS_NOT_INITIALIZED when calling 'cublasCreate(handle)' Open

deespeed chat system

deepspeedai/DeepSpeedExamples#499 · 1 comment · 1 assignee · Opened May 8, 2023
deepspeed hybrid-engine support bloom model with zero3? Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#497 · 3 comments · 1 assignee · Opened May 8, 2023
how to use 'load_in_8bit=True' when train Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#494 · 3 comments · 1 assignee · Opened May 8, 2023
The reward in step3 seems to be completely random without any noticeable increase. Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#489 · 10 comments · 1 assignee · Opened May 7, 2023
program stop when i run my run_chinese.sh in GPU A100*4 80G Open

deespeed chat

deepspeedai/DeepSpeedExamples#487 · 3 comments · 1 assignee · Opened May 6, 2023
A100 40 GB: OOM on step-3 for opt-6.7B Open

deespeed chat new-config system

deepspeedai/DeepSpeedExamples#482 · 4 comments · 1 assignee · Opened May 5, 2023
unable to load 4 7b size model in step3 Open

deespeed chat new-config system

deepspeedai/DeepSpeedExamples#480 · 3 comments · 1 assignee · Opened May 5, 2023
Can not use bloom-560m model in the step2_reward_model_finetuning Open

deespeed chat new-config system

deepspeedai/DeepSpeedExamples#479 · 5 comments · 1 assignee · Opened May 5, 2023
inference OOM Open

deespeed chat

deepspeedai/DeepSpeedExamples#478 · 2 comments · 1 reaction · 1 assignee · Opened May 5, 2023
PPO training unable to reproduce the training log provided Open

deespeed chat modeling

deepspeedai/DeepSpeedExamples#474 · 1 comment · 1 assignee · Opened May 4, 2023
Adding two loss from actor will lead to an error " gradient computed twice for this partition" Open

deespeed chat modeling new-config

deepspeedai/DeepSpeedExamples#458 · 4 comments · 1 assignee · Opened Apr 28, 2023
enable_hybrid_engine issue Open

deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#456 · 10 comments · 3 reactions · 1 assignee · Opened Apr 28, 2023
[BUG]RuntimeError: CUDA error: unknown error Open

deespeed chat system

deepspeedai/DeepSpeedExamples#453 · 3 comments · 1 assignee · Opened Apr 28, 2023
Feature Request: add LazyPromptDataset to DeepSpeedChat Open

deespeed chat enhancement

deepspeedai/DeepSpeedExamples#450 · 4 comments · 1 assignee · Opened Apr 27, 2023
[DeepSpeedExamples/applications/DeepSpeed-Chat/] Error happened when running step3_rlhf_finetuning in enable_hybrid_engine mode with togethercomputer/GPT-NeoXT-Chat-Base-20B Open

deespeed chat hybrid engine

deepspeedai/DeepSpeedExamples#448 · 2 comments · 1 assignee · Opened Apr 27, 2023
training 12b model seems to require more memory than expected Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#447 · 2 comments · 2 reactions · 2 assignees · Opened Apr 27, 2023
is there any expample for zero-inference Open

deepspeedai/DeepSpeedExamples#446 · 3 comments · 1 assignee · Opened Apr 27, 2023
How to train deepspeed-chat using nccl with multi-nodes? Open

deespeed chat system

deepspeedai/DeepSpeedExamples#443 · 4 comments · 1 assignee · Opened Apr 27, 2023
How to use data parallelism between nodes? Open

auto-feature-selection

deepspeedai/DeepSpeedExamples#439 · 1 assignee · Opened Apr 26, 2023
gpt ppo training error Open

deespeed chat new-config

deepspeedai/DeepSpeedExamples#435 · 6 comments · 1 assignee · Opened Apr 26, 2023
the DeepSpeed-Chat demo train.py cannot even run Open

deespeed chat system

deepspeedai/DeepSpeedExamples#432 · 3 comments · 1 assignee · Opened Apr 25, 2023
is Deepspeed-Chat support tensor parallelism for Codegen Open

deespeed chat question

deepspeedai/DeepSpeedExamples#431 · 1 assignee · Opened Apr 25, 2023
[ERROR] [launch.py:434:sigkill_handler] Open

deespeed chat system

deepspeedai/DeepSpeedExamples#430 · 2 comments · 1 assignee · Opened Apr 25, 2023

deepspeedai/DeepSpeedExamples

Get new issues in your inbox