Stas Bekman

stas

https://stasosphere.com/machine-learning/

AI & ML interests

Toolmaker. Software creator, optimizer and harmonizer. Makes things work and fly at Snowflake AI Research Training LLM/RAG/Generative AI/Machine Learning/Scalability

Recent Activity

posted an update about 5 hours ago

I present to you a new experimental open book. https://github.com/stas00/python-cookbook I took my dense Python cheatsheet that I have been honing for many years and use a lot daily and turned it into a book of recipes. Is this useful? This is, of course, free, like other open books.

updated a model about 22 hours ago

stas/python-cookbook

published a model about 22 hours ago

stas/python-cookbook

View all activity

Organizations

posted an update about 5 hours ago

Post

I present to you a new experimental open book.

https://github.com/stas00/python-cookbook

I took my dense Python cheatsheet that I have been honing for many years and use a lot daily and turned it into a book of recipes.

Is this useful?

This is, of course, free, like other open books.

replied to their post about 23 hours ago

I hear you. I suppose I have never been in a situation where I continued using the same debug code for months in a row, usually something is broken, I diagnose writing appropriate tools, fix it and move on. The next time the situation is different, so I always adapted any previous tooling.

Thank you for the kind appreciation and supporting me limiting the scope to what I know, Dipankar.

replied to their post about 24 hours ago

I'm not sure, In my book I share practices that work for me, I have never had a "dull" tool in my debugging practice thus it's not there.

replied to their post 1 day ago

I'm new to using AI so I don't yet have enough experience to answer your question.

Until now all the debugging I did was done manually so nothing was aged or silently dropped, haven't had this problem.

I imagine that with time there will be new sections covering AI-use specifics, but I need to gain more experience first.

posted an update 1 day ago

Post

The Art of Debugging Open Free book is now available in pdf/epub and finally sports a book cover

https://github.com/stas00/the-art-of-debugging#ebook-versions-of-the-book

While a lot of the focus is on Unix/Python/Pytorch, the methodology chapter is applicable to any Software Debugging.

It currently sports 161 packed pages in 5 solid chapters and more coming...

7 replies

replied to their post 2 days ago

At the moment it's a an exact-match, later will support partial match as well.

replied to their post 2 days ago

Thank you for the kind words, Dipankar

The lion share of speed up comes from prompt deduplication during generation and training.

posted an update 3 days ago

Post

111

In parallel we announce a new open source repo:

https://github.com/Snowflake-AI-Research/Arctic-Platform

This is the framework for very fast RL (and future other optimizations rolled into it)

It currently has all the code you need to use or integrate Arctic RL into RL frameworks, with SkyRL and Verl available and more framework integrations coming.

Please kindly spread the word! Thank you!

posted an update 3 days ago

Post

3532

After many months of intense work the
Snowflake AI Research team is happy to present to you the new open source project: Arctic RL

https://snowflake.com/en/blog/engineering/arctic-rl-open-source-backend/

- Arctic RL integrates with VeRL and SkyRL today; enable ZoRRo with one config flag, no code changes required
- ZoRRo delivers up to 6x actor-update acceleration and a 3.5x end-to-end training speedup, reducing Arctic-Text2SQL-R2 training from ~5 days to ~36 hours on 32 H200 GPUs
- Arctic-Text2SQL-R2 achieved higher accuracy scores (48.7) than Gemini 3.1 Pro (47.9) and Claude 4.7 (47.3) on Snowflake's evaluated enterprise SQL benchmark under the tested conditions
- Two open source recipes ship with this release: a text-to-SQL recipe that improved BIRD dev accuracy from 59.92% to 70.35%, and a multi-hop QA recipe that improved average accuracy from 69.6% to 72.3%

4 replies

posted an update 16 days ago

Post

137

PSA for DeepSpeed users - a long outstanding precision-related critical bug has been identified and fixed in https://github.com/deepspeedai/DeepSpeed/pull/8066 and a new release has been made.

The issue was about mixed precision mode downcasting buffers that had to be in fp32 - massively impacting correctness due to large static buffers - e.g. RoPE in Qwen3 models when using long sequence lengths 32K+.

Hopefully this fix brings Deepspeed to a close parity with FSDP2 which has been an issue since a long time.

You can still have the old behavior but you'd now need to manually configure it - by default the model's buffers will now remain in the original precision.

Please install deepspeed==0.19.2 which will do the right thing.

Thanks to Tunji Ruwase and Claude Opus 4.8 via Cursor for identifying and fixing the problem.

posted an update 4 months ago

Post

255

Good news! Ulysses Sequence Parallelism from the Snowflake AI Research and the Deepspeed teams has been integrated into
HuggingFace Trainer, Accelerate and TRL

For extensive details please see this writeup:
https://huggingface.co/blog/ulysses-sp

Thanks a lot to Kashif Rasul for helping make it happen. Also the others in the HF team who helped with integration.

posted an update over 1 year ago

Post

2390

Do you want ArcticTraining at @SnowflakeDB to add an ability to post-train DeepSeek V3/R1 models with DPO using just a few GPU nodes?

Please vote here and tell others about it: https://github.com/snowflakedb/ArcticTraining/discussions/58

ArcticTraining is an open-source, easy to use post-training framework for NVIDIA GPUs built on top of DeepSpeed.

posted an update over 1 year ago

Post

1269

If you remember my work on MAMF - to find the realistic TFLOPS achievable ceiling - the Intel AI team has shared their measurements and they scored ...

an incredible 99.4% TFLOPS efficiency for Gaudi 2!

That's quite amazing! Your ROI on these accelerators will be very high.

The full table is here: https://github.com/stas00/ml-engineering/tree/master/compute/accelerator#maximum-achievable-matmul-flops-comparison-table

As we have seen the competitors get their achievable efficiency worse with each new generation, I'm looking forward to see if Gaudi 3 will keep the high bar!

Thanks to Avi Rubin, Lakshman Chari, Imtiaz Sajwani, Ramy J and Zhiqi Tao for helping to get these numbers to the community.

posted an update almost 2 years ago

Post

1200

The Universal Checkpointing paper is out! https://arxiv.org/abs/2406.18820

If you remember the Bigscience BLOOM-176B training, Tunji Ruwase and I co-invented this technology for Megatron-Deepspeed in order to enable to quickly scale up and down node topology while continuing training.

Since then the DeepSpeed team continued improving on that and it has now been fully integrated into Deepspeed.

The blog post is here: https://github.com/microsoft/DeepSpeed/blob/master/blogs/deepspeed-ucp/README.md

reacted to their post with 🤗 over 2 years ago

Post

A combined effort from the IBM + Pytorch teams achieved an incredible training performance with ZeRO/FSDP on par with 3D parallelism on H100s, while having just 800Gbps inter-node connection.

This is because they got an almost full overlap between comms and compute and have introduced a novel selective activation recomputation method which recalculates only large but inexpensive activations.

Check out their post here: https://pytorch.org/blog/maximizing-training/

posted an update over 2 years ago

Post

posted an update over 2 years ago

Post

If you're trying to run MoE Mixtral-8x7b under DeepSpeed w/ HF Transformers it's likely to hang on the first forward.

The solution is here https://github.com/microsoft/DeepSpeed/pull/4966?_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en-US#issuecomment-1989671378

and you need deepspeed>=0.13.0

Thanks to Masahiro Tanaka for the fix.

replied to their post over 2 years ago

I pinged Elio to see if he wants to join.

posted an update over 2 years ago

Post

Hear, hear, AMD MI300Xs have started to emerge much sooner than expected.

Here is a 2-part benchmarks report on performing BLOOM-176B inference using @MSFTDeepSpeed optimized for AMD MI300X.

1. https://www.evp.cloud/post/diving-deeper-insights-from-our-llm-inference-testing
2. https://www.evp.cloud/post/diving-deeper-insights-from-our-llm-inference-testing-part-2

This was published in response to our BLOOM-176B super-fast inference blog post https://huggingface.co/blog/bloom-inference-pytorch-scripts

Note that these have 192GB of HBM!

The NVIDIA monopoly is strong, but it'll have to start sharing the pie and hopefully drive the costs down at least somewhat.

Thanks to https://www.linkedin.com/in/eliovp for sharing this writeup with me.

p.s. at the PyTorch conference in the fall, the AMD representative said we will see MI300X available to us mortals in Q4-2024/Q1-2025.

5 replies

replied to their post over 2 years ago

Thank you for the kind words, Jeff!

We are still waiting for BLOOM v2.0 from HF!

Stas Bekman

AI & ML interests

Recent Activity

Organizations

stas's activity