Pinned
Paul Chang
@mummitrollet.bsky.social
· Mar 24
Speaking with @trappmartin.bsky.social, I realised there is no good space online to discuss ML-related stuff. It feels like a lot of people are on Bsky, but not much discussion I am myself guilty. So I am going to try and be more active here.
Reposted by Paul Chang
I wrote something about building systems people actually want as an academic in ML. It's pretty much an open letter to 6-months-ago me.
magnusross.github.io/posts/moms-m...
magnusross.github.io/posts/moms-m...
Moms, Models and Medicine | Magnus Ross
A good friend of mine is deep in the world of startups and spends a lot of his time doing idea validation—that is, trying to understand if there is a market for a given idea or product. Despite the fact that in the startup world success is eventually judged by sales or profits, whereas in ML4H it is more likely to be adoption by clinicians and, hopefully, an associated improvement in clinical outcomes, both rely on designing something that people actually want and will use. Therefore, I think many of the tools that help entrepreneurs validate ideas can be repurposed to help researchers undertake projects with real impact.
magnusross.github.io
August 29, 2025 at 1:12 PM
I wrote something about building systems people actually want as an academic in ML. It's pretty much an open letter to 6-months-ago me.
magnusross.github.io/posts/moms-m...
magnusross.github.io/posts/moms-m...
Reposted by Paul Chang
❗️ We just expanded our capacity of B200 SXM6 180GB servers – available in the DataCrunch Cloud Platform.
The best thing is…
You can deploy the Blackwell platform without approvals.
Just sign in, select the instance type, and start your deployment:
cloud.datacrunch.io?utm_source=b...
The best thing is…
You can deploy the Blackwell platform without approvals.
Just sign in, select the instance type, and start your deployment:
cloud.datacrunch.io?utm_source=b...
June 25, 2025 at 5:40 PM
❗️ We just expanded our capacity of B200 SXM6 180GB servers – available in the DataCrunch Cloud Platform.
The best thing is…
You can deploy the Blackwell platform without approvals.
Just sign in, select the instance type, and start your deployment:
cloud.datacrunch.io?utm_source=b...
The best thing is…
You can deploy the Blackwell platform without approvals.
Just sign in, select the instance type, and start your deployment:
cloud.datacrunch.io?utm_source=b...
Reposted by Paul Chang
🆕 Inference API for FLUX.1 Kontext [max] & [pro] are now available on DataCrunch!
We are an infrastructure partner of Black Forest Labs for Kontext, a suite of generative flow matching models for text-to-image and image-to-image editing.
Learn more: datacrunch.io/managed-endp...
We are an infrastructure partner of Black Forest Labs for Kontext, a suite of generative flow matching models for text-to-image and image-to-image editing.
Learn more: datacrunch.io/managed-endp...
May 29, 2025 at 8:51 PM
🆕 Inference API for FLUX.1 Kontext [max] & [pro] are now available on DataCrunch!
We are an infrastructure partner of Black Forest Labs for Kontext, a suite of generative flow matching models for text-to-image and image-to-image editing.
Learn more: datacrunch.io/managed-endp...
We are an infrastructure partner of Black Forest Labs for Kontext, a suite of generative flow matching models for text-to-image and image-to-image editing.
Learn more: datacrunch.io/managed-endp...
Reposted by Paul Chang
🚨 Summer Inference by Symposium AI is happening next Wednesday, June 4, at 16:00-22:00.
🇫🇮 This event will bring together 250 AI engineers, researchers, and founders under one roof in Helsinki.
🔗 You can still grab one of the last remaining seats: lu.ma/x5hhj79x
🇫🇮 This event will bring together 250 AI engineers, researchers, and founders under one roof in Helsinki.
🔗 You can still grab one of the last remaining seats: lu.ma/x5hhj79x
Symposium AI - Summer Inference · Luma
Join 250 leading AI builders for an epic night in Helsinki!
Symposium AI events bring together top AI talent, researchers, and engineers who are actively…
lu.ma
May 26, 2025 at 1:37 PM
🚨 Summer Inference by Symposium AI is happening next Wednesday, June 4, at 16:00-22:00.
🇫🇮 This event will bring together 250 AI engineers, researchers, and founders under one roof in Helsinki.
🔗 You can still grab one of the last remaining seats: lu.ma/x5hhj79x
🇫🇮 This event will bring together 250 AI engineers, researchers, and founders under one roof in Helsinki.
🔗 You can still grab one of the last remaining seats: lu.ma/x5hhj79x
Algorithm hardware co-design was a big reason the whale 🐋(DeepSeek) made such a splash 💦 with its V3 and R1 releases.
May 9, 2025 at 7:52 AM
Algorithm hardware co-design was a big reason the whale 🐋(DeepSeek) made such a splash 💦 with its V3 and R1 releases.
Reposted by Paul Chang
"Cost-aware simulation-based inference" is accepted at AISTATS 2025.
Check out our poster #205 on Sunday May 4th in Hall A-E if you are in Phuket. Finland's rising star @huangdaolang.bsky.social will be there to assist you :D
arxiv.org/abs/2410.07930
@fxbriol.bsky.social @samikaski.bsky.social
Check out our poster #205 on Sunday May 4th in Hall A-E if you are in Phuket. Finland's rising star @huangdaolang.bsky.social will be there to assist you :D
arxiv.org/abs/2410.07930
@fxbriol.bsky.social @samikaski.bsky.social
Cost-aware simulation-based inference
Simulation-based inference (SBI) is the preferred framework for estimating parameters of intractable models in science and engineering. A significant challenge in this context is the large computation...
arxiv.org
May 2, 2025 at 6:45 AM
"Cost-aware simulation-based inference" is accepted at AISTATS 2025.
Check out our poster #205 on Sunday May 4th in Hall A-E if you are in Phuket. Finland's rising star @huangdaolang.bsky.social will be there to assist you :D
arxiv.org/abs/2410.07930
@fxbriol.bsky.social @samikaski.bsky.social
Check out our poster #205 on Sunday May 4th in Hall A-E if you are in Phuket. Finland's rising star @huangdaolang.bsky.social will be there to assist you :D
arxiv.org/abs/2410.07930
@fxbriol.bsky.social @samikaski.bsky.social
Reposted by Paul Chang
I don’t mean to be a broken record but AI development could stop at the o3/Gemini 2.5 level and we would have a decade of major changes across entire professions & industries (medicine, law, education, coding…) as we figure out how to actually use it & adapt our systems.
AI disruption is baked in.
AI disruption is baked in.
April 26, 2025 at 9:57 PM
I don’t mean to be a broken record but AI development could stop at the o3/Gemini 2.5 level and we would have a decade of major changes across entire professions & industries (medicine, law, education, coding…) as we figure out how to actually use it & adapt our systems.
AI disruption is baked in.
AI disruption is baked in.
Reposted by Paul Chang
1/ If you are at ICLR / AABI / AISTATS, check out work from our lab and collaborators on *inference everywhere anytime all at once*!
Go talk to my incredible PhD students @huangdaolang.bsky.social & @chengkunli.bsky.social + amazing collaborator Severi Rissanen.
@univhelsinkics.bsky.social FCAI
Go talk to my incredible PhD students @huangdaolang.bsky.social & @chengkunli.bsky.social + amazing collaborator Severi Rissanen.
@univhelsinkics.bsky.social FCAI
April 27, 2025 at 8:53 AM
1/ If you are at ICLR / AABI / AISTATS, check out work from our lab and collaborators on *inference everywhere anytime all at once*!
Go talk to my incredible PhD students @huangdaolang.bsky.social & @chengkunli.bsky.social + amazing collaborator Severi Rissanen.
@univhelsinkics.bsky.social FCAI
Go talk to my incredible PhD students @huangdaolang.bsky.social & @chengkunli.bsky.social + amazing collaborator Severi Rissanen.
@univhelsinkics.bsky.social FCAI
Reposted by Paul Chang
I wrote something up for AI people who want to get into bluesky and either couldn't assemble an exciting feed or gave up doomscrolling when their Following feed switched to talking politics 24/7.
The AI Researcher's Guide to a Non-Boring Bluesky Feed | Naomi Saphra
How to migrate to bsky without a boring feed.
nsaphra.net
April 26, 2025 at 1:31 AM
I wrote something up for AI people who want to get into bluesky and either couldn't assemble an exciting feed or gave up doomscrolling when their Following feed switched to talking politics 24/7.
Reposted by Paul Chang
1/10🔥 New paper alert in #AABI2025 Proceedings!
Normalizing Flow Regression (NFR) — an offline Bayesian inference method.
What if you could get a full posterior using *only* the evaluations you *already* have, maybe from optimization runs?
Normalizing Flow Regression (NFR) — an offline Bayesian inference method.
What if you could get a full posterior using *only* the evaluations you *already* have, maybe from optimization runs?
April 22, 2025 at 11:01 AM
1/10🔥 New paper alert in #AABI2025 Proceedings!
Normalizing Flow Regression (NFR) — an offline Bayesian inference method.
What if you could get a full posterior using *only* the evaluations you *already* have, maybe from optimization runs?
Normalizing Flow Regression (NFR) — an offline Bayesian inference method.
What if you could get a full posterior using *only* the evaluations you *already* have, maybe from optimization runs?
Reposted by Paul Chang
Tired of your open-source ML work not getting the academic recognition it deserves? 🤔 Submit to the first-ever CodeML workshop at #ICML2025! It focuses on new libraries, improvements to established ones, best practices, retrospectives, and more.
codeml-workshop.github.io/codeml2025/
codeml-workshop.github.io/codeml2025/
CODEML Workshop
Championing Open-source Development in Machine Learning.
codeml-workshop.github.io
April 16, 2025 at 10:15 AM
Tired of your open-source ML work not getting the academic recognition it deserves? 🤔 Submit to the first-ever CodeML workshop at #ICML2025! It focuses on new libraries, improvements to established ones, best practices, retrospectives, and more.
codeml-workshop.github.io/codeml2025/
codeml-workshop.github.io/codeml2025/
Reposted by Paul Chang
This is a great list, things that “the best engineers I know” do, stuff like:
- understanding things deeply, reading the actual source
- being willing to help other people
- status doesn’t matter, good ideas come from anywhere
endler.dev/2025/best-pr...
- understanding things deeply, reading the actual source
- being willing to help other people
- status doesn’t matter, good ideas come from anywhere
endler.dev/2025/best-pr...
The Best Programmers I Know | Matthias Endler
I have met a lot of developers in my life.
Late…
endler.dev
April 13, 2025 at 3:57 PM
This is a great list, things that “the best engineers I know” do, stuff like:
- understanding things deeply, reading the actual source
- being willing to help other people
- status doesn’t matter, good ideas come from anywhere
endler.dev/2025/best-pr...
- understanding things deeply, reading the actual source
- being willing to help other people
- status doesn’t matter, good ideas come from anywhere
endler.dev/2025/best-pr...
B200 go brrrr! It seems by doubling the TFLOPs you get double the speed.
Cool stuff by Antonio and WavespeedAI team to get FLUX-dev inference (SOTA diffusion) in under a second on a B200.
datacrunch.io/blog/flux-on...
Cool stuff by Antonio and WavespeedAI team to get FLUX-dev inference (SOTA diffusion) in under a second on a B200.
datacrunch.io/blog/flux-on...
FLUX on B200: Real-Time Image Inference with WaveSpeedAI + DataCrunch Collaboration
How WaveSpeedAI and DataCrunch achieved an up to 6x faster image inference by optimizing FLUX-dev's latency and efficiency: NVIDIA B200 vs. H100 benchmark.
datacrunch.io
April 9, 2025 at 5:42 AM
B200 go brrrr! It seems by doubling the TFLOPs you get double the speed.
Cool stuff by Antonio and WavespeedAI team to get FLUX-dev inference (SOTA diffusion) in under a second on a B200.
datacrunch.io/blog/flux-on...
Cool stuff by Antonio and WavespeedAI team to get FLUX-dev inference (SOTA diffusion) in under a second on a B200.
datacrunch.io/blog/flux-on...
Reposted by Paul Chang
1/ We asked GPT-4.5 -- allegedly the model with the best sense of humor, according to the site we do not talk about here -- to write a comic about our recent AISTATS paper on the Amortized Conditioning Engine (ACE). Then gpt-4o drew it.
You judge the result...
(text continues 👇)
You judge the result...
(text continues 👇)
April 8, 2025 at 3:28 PM
1/ We asked GPT-4.5 -- allegedly the model with the best sense of humor, according to the site we do not talk about here -- to write a comic about our recent AISTATS paper on the Amortized Conditioning Engine (ACE). Then gpt-4o drew it.
You judge the result...
(text continues 👇)
You judge the result...
(text continues 👇)
Reposted by Paul Chang
The Llama 4 model that won in LM Arena is different than the released version. I have been comparing the answers from Arena to the released model. They aren't close.
The data is worth a look also as it shows how LM Arena results can be manipulated to be more pleasing to humans. t.co/rqAey9SMwh
The data is worth a look also as it shows how LM Arena results can be manipulated to be more pleasing to humans. t.co/rqAey9SMwh
April 8, 2025 at 2:10 AM
The Llama 4 model that won in LM Arena is different than the released version. I have been comparing the answers from Arena to the released model. They aren't close.
The data is worth a look also as it shows how LM Arena results can be manipulated to be more pleasing to humans. t.co/rqAey9SMwh
The data is worth a look also as it shows how LM Arena results can be manipulated to be more pleasing to humans. t.co/rqAey9SMwh
I wanted to change the color scheme on a blog for some plots so I decided to test Claude code.
github.com/datacrunch-r...
I didnt relaize it inserts
"Co-Authored-By: Claude"
I would have got away with it if it wasn't for that pesky Claude code.
github.com/datacrunch-r...
I didnt relaize it inserts
"Co-Authored-By: Claude
I would have got away with it if it wasn't for that pesky Claude code.
Update plot-script.py to use 2025 brand colors · datacrunch-research/blogs@6955de0
- Added brand colors 2025 palette
- Updated all plots to use the new color scheme
- Regenerated all plot images with the new colors
- Set consistent style across all plots
🤖 Generated with [Claude...
github.com
April 8, 2025 at 8:50 AM
I wanted to change the color scheme on a blog for some plots so I decided to test Claude code.
github.com/datacrunch-r...
I didnt relaize it inserts
"Co-Authored-By: Claude"
I would have got away with it if it wasn't for that pesky Claude code.
github.com/datacrunch-r...
I didnt relaize it inserts
"Co-Authored-By: Claude
I would have got away with it if it wasn't for that pesky Claude code.
This is a new blog looking at the individual optimizations that went into serving the DeepSeek model class in SGLang. I have been observing the SGLang repo for a few months now, and it's crazy how quickly they integrate new optimized features. It's a very cool open-source project!
New blog post: Optimization techniques applied by the SGLang team for DeepSeek-V3 inference.
You'll find a comprehensive overview of the techniques, their benefits and implications, and our benchmarks.
datacrunch.io/blog/deepsee...
You'll find a comprehensive overview of the techniques, their benefits and implications, and our benchmarks.
datacrunch.io/blog/deepsee...
DeepSeek-V3 + SGLang: Inference Optimization
A comprehensive overview of optimization techniques applied by the SGLang team for DeepSeek-V3 inference with GitHub commits, benchmarks, and results.
datacrunch.io
April 8, 2025 at 7:39 AM
This is a new blog looking at the individual optimizations that went into serving the DeepSeek model class in SGLang. I have been observing the SGLang repo for a few months now, and it's crazy how quickly they integrate new optimized features. It's a very cool open-source project!
Llama 4 uses both interleaved chunked attention and global (NoPE) attention mechanisms, similar to a recent Cohere paper. It's cool to see innovation in attention layer architectures for the large models, and it showed to the world.
arxiv.org/abs/2501.18795.
arxiv.org/abs/2501.18795.
Rope to Nope and Back Again: A New Hybrid Attention Strategy
Long-context large language models (LLMs) have achieved remarkable advancements, driven by techniques like Rotary Position Embedding (RoPE) (Su et al., 2023) and its extensions (Chen et al., 2023; Liu...
arxiv.org
April 6, 2025 at 7:20 AM
Llama 4 uses both interleaved chunked attention and global (NoPE) attention mechanisms, similar to a recent Cohere paper. It's cool to see innovation in attention layer architectures for the large models, and it showed to the world.
arxiv.org/abs/2501.18795.
arxiv.org/abs/2501.18795.
Reposted by Paul Chang
The Johnson–Lindenstrauss Lemma states that a set of high-dimensional points can be mapped into a much lower-dimensional space while approximately preserving pairwise distances. This is useful for dimensionality reduction, clustering, etc. stanford.edu/class/cs114/...
April 4, 2025 at 5:00 AM
The Johnson–Lindenstrauss Lemma states that a set of high-dimensional points can be mapped into a much lower-dimensional space while approximately preserving pairwise distances. This is useful for dimensionality reduction, clustering, etc. stanford.edu/class/cs114/...
WaveSpeedAI Docs
Ultimate APl for Accelerating Al Image and Video Generation
wavespeed.ai
April 4, 2025 at 11:30 AM
Reposted by Paul Chang
🚨 NVIDIA HGX B200: available NOW on DataCrunch!
Be among the first to gain instant access to 1x, 2x, 4x, and 8x B200 GPUs with our high-performance VMs.
Sign up and enjoy expert support with secure service where performance meets sustainability.
🔗 cloud.datacrunch.io
Be among the first to gain instant access to 1x, 2x, 4x, and 8x B200 GPUs with our high-performance VMs.
Sign up and enjoy expert support with secure service where performance meets sustainability.
🔗 cloud.datacrunch.io
March 31, 2025 at 1:42 PM
🚨 NVIDIA HGX B200: available NOW on DataCrunch!
Be among the first to gain instant access to 1x, 2x, 4x, and 8x B200 GPUs with our high-performance VMs.
Sign up and enjoy expert support with secure service where performance meets sustainability.
🔗 cloud.datacrunch.io
Be among the first to gain instant access to 1x, 2x, 4x, and 8x B200 GPUs with our high-performance VMs.
Sign up and enjoy expert support with secure service where performance meets sustainability.
🔗 cloud.datacrunch.io
Reposted by Paul Chang
I just came across this paper from ICLR 2024 which proposes an intricate combination of transformers and diffusion models to generate forecasts with these uncertainty bounds (red box), which are clearly inappropriate and could likely be outperformed by modelling the data as a random walk...
March 31, 2025 at 8:41 AM
I just came across this paper from ICLR 2024 which proposes an intricate combination of transformers and diffusion models to generate forecasts with these uncertainty bounds (red box), which are clearly inappropriate and could likely be outperformed by modelling the data as a random walk...
Reposted by Paul Chang
🥉 SemiAnalysis awarded DataCrunch with bronze on the GPU Cloud ClusterMAX™ Rating!
We thank their team for this independent evaluation, validating our approach to pushing the boundary of resource-efficient AI infrastructure ⬇️
We thank their team for this independent evaluation, validating our approach to pushing the boundary of resource-efficient AI infrastructure ⬇️
March 28, 2025 at 8:45 PM
🥉 SemiAnalysis awarded DataCrunch with bronze on the GPU Cloud ClusterMAX™ Rating!
We thank their team for this independent evaluation, validating our approach to pushing the boundary of resource-efficient AI infrastructure ⬇️
We thank their team for this independent evaluation, validating our approach to pushing the boundary of resource-efficient AI infrastructure ⬇️
Pretty exciting idea about enforcing sparsity by changing the activation function used.
arxiv.org/abs/2503.16672
arxiv.org/abs/2503.16672
Accelerating Transformer Inference and Training with 2:4 Activation Sparsity
In this paper, we demonstrate how to leverage 2:4 sparsity, a popular hardware-accelerated GPU sparsity pattern, to activations to accelerate large language model training and inference. Crucially we ...
arxiv.org
March 28, 2025 at 6:59 AM
Pretty exciting idea about enforcing sparsity by changing the activation function used.
arxiv.org/abs/2503.16672
arxiv.org/abs/2503.16672