Picture

DeepSpeed @DeepSpeedAI

Official account for DeepSpeed, a library that enables unprecedented scale and speed for deep learning training + inference. 日本語 : @DeepSpeedAI_JP deepspeed.ai Joined May 2020
  • Tweets
    88
  • Followers
    4K
  • Following
    88
  • Likes
    84
DeepSpeed retweeted
StasBekman Profile Picture

Stas Bekman @StasBekman

2 weeks ago

The @DeepSpeedAI would like to thank @modal for sponsoring our gpus for CI. This is an amazing contribution to our AI-democratizing open source project. github.com/deepspeedai/De… The Modal team is outstanding in their amazing support - speed, expertise and a human experience!

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

2 months ago

Kudos to Xinyu for giving an excellent presentation of DeepSpeed Universal Checkpointing (UCP) paper at USENIX ATC 2015.

_Minjia_Zhang_ Profile Picture

Minjia Zhang @_Minjia_Zhang_

2 months ago

Kudos to Xinyu for giving an excellent presentation of DeepSpeed Universal Checkpointing (UCP) paper at USENIX ATC 2015.

DeepSpeed retweeted
StasBekman Profile Picture

Stas Bekman @StasBekman

2 months ago

My first project at @Snowflake AI Research is complete! I present to you Arctic Long Sequence Training (ALST) Paper: arxiv.org/abs/2506.13996 Blog: snowflake.com/en/engineering… ALST is a set of modular, open-source techniques that enable training on sequences up to 15 million…

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

3 months ago

Improved DeepNVMe: Affordable I/O Scaling for AI - Faster I/O with PCIe Gen5 - 20x faster model checkpointing - Low-budget SGLang inference via NVMe offloading - Pinned memory for CPU-only workloads - Zero-copy tensor type casting Blog: tinyurl.com/yanbrjy9

DeepSpeed retweeted
PyTorch Profile Picture

PyTorch @PyTorch

4 months ago

PyTorch Foundation has expanded into an umbrella foundation. @vllm_project and @DeepSpeedAI have been accepted as hosted projects, advancing community-driven AI across the full lifecycle. Supporting quotes provided by the following members: @AMD, @Arm, @AWS, @Google, @Huawei,…

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

4 months ago

Come hear all the exciting DeepSpeed updates at the upcoming PyTorch Day France 2025 DeepSpeed – Efficient Training Scalability for Deep Learning Models - sched.co/21nyy @sched

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

5 months ago

Introducing 🚀DeepCompile🚀: compiler-based distributed training optimizations. - Automatic parallelization & profile-guided optimizations - Enable ZeRO1, ZeRO3, Offloading, etc. via compiler passes - 1.2X-7X speedups over manual ZeRO1/ZeRO3/Offloading tinyurl.com/8cys28xk

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

5 months ago

AutoTP + ZeRO Training for HF Models - Enhance HF post-training with larger models, batches, & contexts - 4x faster LLAMA3 fine-tuning with TP=2 vs TP=1 - No code changes needed Blog: tinyurl.com/5n8nfs2w

DeepSpeed retweeted
xariusrke Profile Picture

xr-5 🐀 @xariusrke

6 months ago

1/4⚡️nanoton now supports DoMiNo with intra-layer communication overlapping, achieving 60% communication hiding for tensor parallelism (TP) in both the forward and backward passes while maintaining the same training loss.

DeepSpeed retweeted
LFAIDataFdn Profile Picture

LF AI & Data Foundation @LFAIDataFdn

7 months ago

🚀 Excited to introduce DeepSpeed, a deep learning optimization library from @Microsoft! It simplifies distributed training and inference, making AI scaling more efficient and cost-effective. Learn more 👉 hubs.la/Q0351DJC0 #DeepSpeed #AI #OpenSource #LFAIData

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

9 months ago

🚀Introducing Ulysses-Offload🚀 - Unlock the power of long context LLM training and finetuning with our latest system optimizations - Train LLaMA3-8B on 2M tokens context using 4xA100-80GB - Achieve over 55% MFU Blog: shorturl.at/Spx6Y Tutorial: shorturl.at/bAWu5

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

9 months ago

Introducing Domino: a novel zero-cost communication tensor parallelism (TP) training engine for both single node and multi-node settings. - Near-complete communication hiding - Novel multi-node scalable TP solution Blog: github.com/microsoft/Deep…

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

a year ago

Great to see the amazing DeepSpeed optimizations from @Guanhua_Wang_, Heyang Qin, @toh_tana, @QuentinAnthon15, and @samadejacobs presented by @ammar_awan at MUG '24.

mvapich Profile Picture

MVAPICH @mvapich

a year ago

Great to see the amazing DeepSpeed optimizations from @Guanhua_Wang_, Heyang Qin, @toh_tana, @QuentinAnthon15, and @samadejacobs presented by @ammar_awan at MUG '24.

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

a year ago

Announcing that DeepSpeed now runs natively on Windows. This exciting combination unlocks DeepSpeed optimizations to Windows users and empowers more people and organizations with AI innovations. - HF Inference & Finetuning - LoRA - CPU Offload Blog: shorturl.at/a7TF8

DeepSpeed retweeted
Cometml Profile Picture

Comet @Cometml

a year ago

💡Check out Comet’s latest integration with DeepSpeed, a deep learning optimization library! 🤝With the @MSFTDeepSpeed + @Cometml integration automatically start logging training metrics generated by DeepSpeed. Try the quick-start Colab to get started: colab.research.google.com/github/comet-m…

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

a year ago

Introducing DeepNVMe, a suite of optimizations for fast and efficient I/O operations in DL applications. - POSIX-style APIs - Direct HBM/NVMe xfers via NVIDIA GDS - Cheap Inference scaling via NVMe-Offload Blog: shorturl.at/l7Oue @Azure @NVIDIADC #FMS24 #GPUDirect

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

a year ago

Introducing Universal Checkpointing for boosting training efficiency. - Change parallelism (PP, SP, TP, ZeRO-DP) or GPU count mid-stream - Improve resilience by scaling down to healthy nodes💪 - Increase throughput by scaling up to elastic nodes🚀 Blog: rb.gy/aup3pn

DeepSpeedAI Profile Picture

DeepSpeed @DeepSpeedAI

2 years ago

#DeepSpeed joins forces with @Sydney_Uni to unveil an exciting tech #FP6. Just supply your FP16 models, and we deliver: 🚀 1.5x performance boost for #LLMs serving on #GPUs 🚀 Innovative (4+2)-bit system design 🚀 Quality-preserving quantization link: github.com/microsoft/Deep…

DeepSpeed retweeted
cloneofsimo Profile Picture

Simo Ryu @cloneofsimo

2 years ago

So you've had your fun with @karpathy 's mingpt. Now its time to scale : introducing min-max-gpt: really small codebase that scales with help of @MSFTDeepSpeed . No huggingface accelerate, transformer. Just deepspeed + torch: maximum hackability github.com/cloneofsimo/mi…

Trends for United States

#UFCParis
58,4 B posts
Troy
14,4 B posts
Staub
3.653 posts
Baylor
12,8 B posts
#USMNT
1.410 posts
Ruffy
27,9 B posts
Stoops
2.306 posts
Kentucky
12,4 B posts
Calzada
2.563 posts
Karen
263 B posts
Oklahoma State
4.999 posts
Oregon
35,8 B posts
Iowa
34,9 B posts
Gundy
2.035 posts
Julian Sayin
2.777 posts
Salter
1.780 posts
Ole Miss
5.880 posts
Billy Napier
Imavov
20,2 B posts
Paul Craig
3.140 posts

You might like

An AI research and product company 🫠. We are a team of scientists and engineers building state-of-the-art multimodal models 😻
Reka
@RekaAILabs
17K Followers
LMArena: Open Platform for Community-driven AI Benchmarking. Graduated from UC Berkeley / @lmsysorg. We’re hiring: https://t.co/1OkfLq2n0I
lmarena.ai
@lmarena_ai
92K Followers
ML Researcher @AnthropicAI. Previously OpenAI & DeepMind. Optimizing for a post-AGI future where humanity flourishes. Opinions aren't my employer's.
Jan Leike
@janleike
114K Followers
Asst. Prof @PrincetonCS, Chief Scientist @togethercompute. Machine learning & systems.
Tri Dao
@tri_dao
32K Followers
We remove the barriers to state-of-the-art generative AI model development and make data + AI available to all.
Databricks Mosaic Res...
@DbrxMosaicAI
41K Followers
@ Google Deepmind. Past: @MetaAI, @OpenAI, @unitygames, @losalamosnatlab, @Princeton etc. Always hungry for intelligence.
Susan Zhang
@suchenzang
33K Followers
Research Scientist @OpenAI Previously: PhD @MIT @MSFTResearch @UberATG @SCSatCMU @AUB_Lebanon
Hadi Salman
@hadisalmanX
6K Followers
@LangChainAI, previously @robusthq @kensho MLOps ∪ Generative AI ∪ sports analytics
Harrison Chase
@hwchase17
79K Followers
co-founder/CEO @llama_index Careers: https://t.co/EUnMNmbCtx Enterprise: https://t.co/Ht5jwxSrQB
Jerry Liu
@jerryjliu0
65K Followers
AI researcher. Interested in Reasoning, Multimodal. I direct TIGER-Lab. Author of PoT, MMMU, MMLU-Pro, MAmmoTH, LongRAG, MAP-Neo, YuE, VL-Rethinker
Wenhu Chen
@WenhuChen
22K Followers
Build AI agents over your documents Github: https://t.co/HC19j7vMwc Docs: https://t.co/QInqg2zksh LlamaCloud: https://t.co/yQGTiRSNvj
LlamaIndex 🦙
@llama_index
101K Followers
research scientist @googledeepmind ✨♊, model co-lead/captain of gemini deepthink imo gold medal 🥇, opinions are my own.
Yi Tay
@YiTayML
46K Followers
24th at the Electrica puzzle challenge | https://t.co/baTQS2bdia
Georgi Gerganov
@ggerganov
52K Followers
Co-founder and CTO at @huggingface 🤗. ML/AI for everyone, building products to propel communities fwd. @Stanford + @Polytechnique
Julien Chaumond
@julien_c
64K Followers
Covering the latest in AI development • ML Eng since 2017 • Building @AlphaSignalAI into the #1 source of news for AI devs → At 250k users.
Lior⚡
@LiorOnAI
106K Followers