Open R1: Update #3
•
225
Open R1
Enterprise
community
AI & ML interests
None defined yet.
Recent Activity
View all activity
Articles
open-r1's activity
lewtun
updated
a
Space
about 9 hours ago
Post
845
smolagents now support vLLM! 🥳
As one of the most popular local inference solutions, the community had been asking us to integrate vLLM: after a heavy refactoring of our LLM classes, we've just released smolagents 1.11.0, with a brand new VLLMModel class.
Go try it and tell us what you think!
https://github.com/huggingface/smolagents/blob/45b2c86857b7f7657daaa74e4d17d347e9e2c4a4/src/smolagents/models.py#L497
As one of the most popular local inference solutions, the community had been asking us to integrate vLLM: after a heavy refactoring of our LLM classes, we've just released smolagents 1.11.0, with a brand new VLLMModel class.
Go try it and tell us what you think!
https://github.com/huggingface/smolagents/blob/45b2c86857b7f7657daaa74e4d17d347e9e2c4a4/src/smolagents/models.py#L497
burtenshaw
posted
an
update
2 days ago
Post
1646
The open LLM leaderboard is completed, retired, dead, ‘ascended to a higher plane’. And in its shadow we have an amazing range of leaderboards built and maintained by the community.
In this post, I just want to list some of those great leaderboards that you should bookmark for staying up to date:
- Chatbot Arena LLM Leaderboard is the first port of call for checking out the best model. It’s not the fastest because humans will need to use the models to get scores, but it’s worth the wait. lmarena-ai/chatbot-arena-leaderboard
- OpenVLM Leaderboard is great for getting scores on vision language models opencompass/open_vlm_leaderboard
- Ai2 are doing a great job on RewardBench and I hope they keep it up because reward models are the unsexy workhorse of the field. allenai/reward-bench
- The GAIA leaderboard is great for evaluating agent applications. gaia-benchmark/leaderboard
🤩 This seems like such a sustainable way of building for the long term, where rather than leaning on a single company to evaluate all LLMs, we share the load.
In this post, I just want to list some of those great leaderboards that you should bookmark for staying up to date:
- Chatbot Arena LLM Leaderboard is the first port of call for checking out the best model. It’s not the fastest because humans will need to use the models to get scores, but it’s worth the wait. lmarena-ai/chatbot-arena-leaderboard
- OpenVLM Leaderboard is great for getting scores on vision language models opencompass/open_vlm_leaderboard
- Ai2 are doing a great job on RewardBench and I hope they keep it up because reward models are the unsexy workhorse of the field. allenai/reward-bench
- The GAIA leaderboard is great for evaluating agent applications. gaia-benchmark/leaderboard
🤩 This seems like such a sustainable way of building for the long term, where rather than leaning on a single company to evaluate all LLMs, we share the load.
- 3 replies
Post
519
🤯 Gemma 3's image analysis blew me away!
Tested 2 ways to extract airplane registration numbers from photos with 12B model:
1️⃣ Gradio app w/API link (underrated feature IMO) + ZeroGPU infra on Hugging Face in Google Colab. Fast & free.
2️⃣ LMStudio + local processing (100% private). Running this powerhouse on a MacBook w/16GB RAM is wild! 🚀
Colab: https://colab.research.google.com/drive/1YmmaP0IDEu98CLDppAAK9kbQZ7lFnLZ1?usp=sharing
Tested 2 ways to extract airplane registration numbers from photos with 12B model:
1️⃣ Gradio app w/API link (underrated feature IMO) + ZeroGPU infra on Hugging Face in Google Colab. Fast & free.
2️⃣ LMStudio + local processing (100% private). Running this powerhouse on a MacBook w/16GB RAM is wild! 🚀
Colab: https://colab.research.google.com/drive/1YmmaP0IDEu98CLDppAAK9kbQZ7lFnLZ1?usp=sharing
burtenshaw
posted
an
update
3 days ago
Post
1651
Still speed running Gemma 3 to think. Today I focused on setting up gpu poor hardware to run GRPO.
This is a plain TRL and PEFT notebook which works on mac silicone or colab T4. This uses the 1b variant of Gemma 3 and a reasoning version of GSM8K dataset.
🧑🍳 There’s more still in the oven like releasing models, an Unsloth version, and deeper tutorials, but hopefully this should bootstrap your projects.
Here’s a link to the 1b notebook: https://colab.research.google.com/drive/1mwCy5GQb9xJFSuwt2L_We3eKkVbx2qSt?usp=sharing
This is a plain TRL and PEFT notebook which works on mac silicone or colab T4. This uses the 1b variant of Gemma 3 and a reasoning version of GSM8K dataset.
🧑🍳 There’s more still in the oven like releasing models, an Unsloth version, and deeper tutorials, but hopefully this should bootstrap your projects.
Here’s a link to the 1b notebook: https://colab.research.google.com/drive/1mwCy5GQb9xJFSuwt2L_We3eKkVbx2qSt?usp=sharing
- 1 reply
burtenshaw
posted
an
update
3 days ago
Post
1546
everybody and their dog is fine-tuning Gemma 3 today, so I thought I'd do a longer post on the tips and sharp edges I find. let's go!
1. has to be install everything form main and nightly. this is what I'm working with to get unsloth and TRL running
plus this with
2. will brown's code to turn GSM8k into a reasoning dataset is a nice toy experiment https://gist.github.com/willccbb/4676755236bb08cab5f4e54a0475d6fb
3. with a learning rate of 5e-6 rewards and loss stayed flat for the first 100 or so steps.
4. so far none of my runs have undermined the outputs after 1 epoch. therefore, I'm mainly experimenting with bigger LoRA adapters.
5. vision fine-tuning isn't available in TRL's GRPOTrainer, so stick to text datasets. but no need to load the model differently in transformers or Unsloth
if you want an introduction to GRPO, check out the reasoning course, it walks you through the algorithm, theory, and implementation in a smooth way.
https://huggingface.co/reasoning-course
1. has to be install everything form main and nightly. this is what I'm working with to get unsloth and TRL running
git+https://github.com/huggingface/transformers@main
git+https://github.com/huggingface/trl.git@main
bitsandbytes
peftplus this with
--no-depsgit+https://github.com/unslothai/unsloth-zoo.git@nightly
git+https://github.com/unslothai/unsloth.git@nightly2. will brown's code to turn GSM8k into a reasoning dataset is a nice toy experiment https://gist.github.com/willccbb/4676755236bb08cab5f4e54a0475d6fb
3. with a learning rate of 5e-6 rewards and loss stayed flat for the first 100 or so steps.
4. so far none of my runs have undermined the outputs after 1 epoch. therefore, I'm mainly experimenting with bigger LoRA adapters.
from trl import GRPOConfig
training_args = GRPOConfig(
learning_rate = 5e-6,
adam_beta1 = 0.9,
adam_beta2 = 0.99,
weight_decay = 0.1,
warmup_ratio = 0.1,
lr_scheduler_type = "cosine",
optim = "adamw_8bit",
logging_steps = 1,
per_device_train_batch_size = 2,
gradient_accumulation_steps = 1,
num_generations = 2,
max_prompt_length = 256,
max_completion_length = 1024 - 256,
num_train_epochs = 1,
max_steps = 250,
save_steps = 250,
max_grad_norm = 0.1,
report_to = "none",
)5. vision fine-tuning isn't available in TRL's GRPOTrainer, so stick to text datasets. but no need to load the model differently in transformers or Unsloth
from transformers import AutoModelForImageTextToText
model = AutoModelForImageTextToText.from_pretrained("google/gemma-3-4b-it)if you want an introduction to GRPO, check out the reasoning course, it walks you through the algorithm, theory, and implementation in a smooth way.
https://huggingface.co/reasoning-course
-
- 2 replies
Post
1149
Ever wanted 45 min with one of AI’s most fascinating minds? Was with
@thomwolf
at HumanX Vegas. Sharing my notes of his Q&A with the press—completely changed how I think about AI’s future:
1️⃣ The next wave of successful AI companies won’t be defined by who has the best model but by who builds the most useful real-world solutions. "We all have engines in our cars, but that’s rarely the only reason we buy one. We expect it to work well, and that’s enough. LLMs will be the same."
2️⃣ Big players are pivoting: "Closed-source companies—OpenAI being the first—have largely shifted from LLM announcements to product announcements."
3️⃣ Open source is changing everything: "DeepSeek was open source AI’s ChatGPT moment. Basically, everyone outside the bubble realized you can get a model for free—and it’s just as good as the paid ones."
4️⃣ Product innovation is being democratized: Take Manus, for example—they built a product on top of Anthropic’s models that’s "actually better than Anthropic’s own product for now, in terms of agents." This proves that anyone can build great products with existing models.
We’re entering a "multi-LLM world," where models are becoming commoditized, and all the tools to build are readily available—just look at the flurry of daily new releases on Hugging Face.
Thom's comparison to the internet era is spot-on: "In the beginning you made a lot of money by making websites... but nowadays the huge internet companies are not the companies that built websites. Like Airbnb, Uber, Facebook, they just use the internet as a medium to make something for real life use cases."
Love to hear your thoughts on this shift!
1️⃣ The next wave of successful AI companies won’t be defined by who has the best model but by who builds the most useful real-world solutions. "We all have engines in our cars, but that’s rarely the only reason we buy one. We expect it to work well, and that’s enough. LLMs will be the same."
2️⃣ Big players are pivoting: "Closed-source companies—OpenAI being the first—have largely shifted from LLM announcements to product announcements."
3️⃣ Open source is changing everything: "DeepSeek was open source AI’s ChatGPT moment. Basically, everyone outside the bubble realized you can get a model for free—and it’s just as good as the paid ones."
4️⃣ Product innovation is being democratized: Take Manus, for example—they built a product on top of Anthropic’s models that’s "actually better than Anthropic’s own product for now, in terms of agents." This proves that anyone can build great products with existing models.
We’re entering a "multi-LLM world," where models are becoming commoditized, and all the tools to build are readily available—just look at the flurry of daily new releases on Hugging Face.
Thom's comparison to the internet era is spot-on: "In the beginning you made a lot of money by making websites... but nowadays the huge internet companies are not the companies that built websites. Like Airbnb, Uber, Facebook, they just use the internet as a medium to make something for real life use cases."
Love to hear your thoughts on this shift!
- 1 reply
burtenshaw
posted
an
update
4 days ago
Post
1699
Here’s a notebook to make Gemma reason with GRPO & TRL. I made this whilst prepping the next unit of the reasoning course:
In this notebooks I combine together google’s model with some community tooling
- First, I load the model from the Hugging Face hub with transformers’s latest release for Gemma 3
- I use PEFT and bitsandbytes to get it running on Colab
- Then, I took Will Browns processing and reward functions to make reasoning chains from GSM8k
- Finally, I used TRL’s GRPOTrainer to train the model
Next step is to bring Unsloth AI in, then ship it in the reasoning course. Links to notebook below.
https://colab.research.google.com/drive/1Vkl69ytCS3bvOtV9_stRETMthlQXR4wX?usp=sharing
In this notebooks I combine together google’s model with some community tooling
- First, I load the model from the Hugging Face hub with transformers’s latest release for Gemma 3
- I use PEFT and bitsandbytes to get it running on Colab
- Then, I took Will Browns processing and reward functions to make reasoning chains from GSM8k
- Finally, I used TRL’s GRPOTrainer to train the model
Next step is to bring Unsloth AI in, then ship it in the reasoning course. Links to notebook below.
https://colab.research.google.com/drive/1Vkl69ytCS3bvOtV9_stRETMthlQXR4wX?usp=sharing
- 3 replies
edbeeching
authored
a
paper
4 days ago
lewtun
authored
a
paper
4 days ago
Post
2127
We've kept pushing our Open-R1 project, an open initiative to replicate and extend the techniques behind DeepSeek-R1.
And even we were mind-blown by the results we got with this latest model we're releasing: ⚡️OlympicCoder ( open-r1/OlympicCoder-7B and open-r1/OlympicCoder-32B)
It's beating Claude 3.7 on (competitive) programming –a domain Anthropic has been historically really strong at– and it's getting close to o1-mini/R1 on olympiad level coding with just 7B parameters!
And the best part is that we're open-sourcing all about its training dataset, the new IOI benchmark, and more in our Open-R1 progress report #3: https://huggingface.co/blog/open-r1/update-3
Datasets are are releasing:
- open-r1/codeforces
- open-r1/codeforces-cots
- open-r1/ioi
- open-r1/ioi-test-cases
- open-r1/ioi-sample-solutions
- open-r1/ioi-cots
- open-r1/ioi-2024-model-solutions
And even we were mind-blown by the results we got with this latest model we're releasing: ⚡️OlympicCoder ( open-r1/OlympicCoder-7B and open-r1/OlympicCoder-32B)
It's beating Claude 3.7 on (competitive) programming –a domain Anthropic has been historically really strong at– and it's getting close to o1-mini/R1 on olympiad level coding with just 7B parameters!
And the best part is that we're open-sourcing all about its training dataset, the new IOI benchmark, and more in our Open-R1 progress report #3: https://huggingface.co/blog/open-r1/update-3
Datasets are are releasing:
- open-r1/codeforces
- open-r1/codeforces-cots
- open-r1/ioi
- open-r1/ioi-test-cases
- open-r1/ioi-sample-solutions
- open-r1/ioi-cots
- open-r1/ioi-2024-model-solutions
Post
1382
Google just dropped an exciting technical report for the brand-new Gemma3 model! 🚀 Here are my personal notes highlighting the most intriguing architectural innovations, design choices, and insights from this release:
1) Architecture choices:
> No more softcaping, replace by QK-Norm
> Both Pre AND Post Norm
> Wider MLP than Qwen2.5, ~ same depth
> SWA with 5:1 and 1024 (very small and cool ablation on the paper!)
> No MLA to save KV cache, SWA do the job!
2) Long context
> Only increase the rope in the global layer (to 1M)
> Confirmation that it's harder to do long context for smol models, no 128k for the 1B
> Pretrained with 32k context? seems very high
> No yarn nor llama3 like rope extension
3) Distillation
> Only keep te first 256 logits for the teacher
> Ablation on the teacher gap (tl;dr you need some "patience" to see that using a small teacher is better)
> On policy distillation yeahh (by
@agarwl_
et al), not sure if the teacher gap behave the same here, curious if someone have more info?
4) Others
> Checkpoint with QAT, that's very cool
> RL using improve version of BOND, WARM/WARP good excuse to look at
@ramealexandre
papers
> Only use Zero3, no TP/PP if i understand correctly ?
> Training budget relatively similar than gemma2
1) Architecture choices:
> No more softcaping, replace by QK-Norm
> Both Pre AND Post Norm
> Wider MLP than Qwen2.5, ~ same depth
> SWA with 5:1 and 1024 (very small and cool ablation on the paper!)
> No MLA to save KV cache, SWA do the job!
2) Long context
> Only increase the rope in the global layer (to 1M)
> Confirmation that it's harder to do long context for smol models, no 128k for the 1B
> Pretrained with 32k context? seems very high
> No yarn nor llama3 like rope extension
3) Distillation
> Only keep te first 256 logits for the teacher
> Ablation on the teacher gap (tl;dr you need some "patience" to see that using a small teacher is better)
> On policy distillation yeahh (by
@agarwl_
et al), not sure if the teacher gap behave the same here, curious if someone have more info?
4) Others
> Checkpoint with QAT, that's very cool
> RL using improve version of BOND, WARM/WARP good excuse to look at
@ramealexandre
papers
> Only use Zero3, no TP/PP if i understand correctly ?
> Training budget relatively similar than gemma2
- 1 reply
clefourrier
posted
an
update
4 days ago
Post
1593
Gemma3 family is out! Reading the tech report, and this section was really interesting to me from a methods/scientific fairness pov.
Instead of doing over-hyped comparisons, they clearly state that **results are reported in a setup which is advantageous to their models**.
(Which everybody does, but people usually don't say)
For a tech report, it makes a lot of sense to report model performance when used optimally!
On leaderboards on the other hand, comparison will be apples to apples, but in a potentially unoptimal way for a given model family (like some user interact sub-optimally with models)
Also contains a cool section (6) on training data memorization rate too! Important to see if your model will output the training data it has seen as such: always an issue for privacy/copyright/... but also very much for evaluation!
Because if your model knows its evals by heart, you're not testing for generalization.
Instead of doing over-hyped comparisons, they clearly state that **results are reported in a setup which is advantageous to their models**.
(Which everybody does, but people usually don't say)
For a tech report, it makes a lot of sense to report model performance when used optimally!
On leaderboards on the other hand, comparison will be apples to apples, but in a potentially unoptimal way for a given model family (like some user interact sub-optimally with models)
Also contains a cool section (6) on training data memorization rate too! Important to see if your model will output the training data it has seen as such: always an issue for privacy/copyright/... but also very much for evaluation!
Because if your model knows its evals by heart, you're not testing for generalization.
Post
1891
Introducing OlympicCoder: a series of open reasoning models that can solve olympiad-level programming problems 🧑💻
- 7B open-r1/OlympicCoder-7B
- 32B open-r1/OlympicCoder-32B
We find that OlympicCoder models outperform Claude 3.7 Sonnet, as well as others over 100x larger 💪
Together with the models, we are releasing:
📊CodeForces-CoTs: new dataset of code problems from the most popular competitive coding platform, with R1 traces in C++ and Python open-r1/codeforces-cots
🏆 IOI'2024: a new benchmark of VERY hard programming problems where even frontier models struggle to match human performance open-r1/ioi
For links to the models and datasets, check out our latest progress report from Open R1: https://huggingface.co/blog/open-r1/update-3
- 7B open-r1/OlympicCoder-7B
- 32B open-r1/OlympicCoder-32B
We find that OlympicCoder models outperform Claude 3.7 Sonnet, as well as others over 100x larger 💪
Together with the models, we are releasing:
📊CodeForces-CoTs: new dataset of code problems from the most popular competitive coding platform, with R1 traces in C++ and Python open-r1/codeforces-cots
🏆 IOI'2024: a new benchmark of VERY hard programming problems where even frontier models struggle to match human performance open-r1/ioi
For links to the models and datasets, check out our latest progress report from Open R1: https://huggingface.co/blog/open-r1/update-3
-
- 1 reply
Post
1702
🔥The Open R1 team just dropped OlympicCoder and it's wild:
- 7B model outperforms Claude 3.7 Sonnet on IOI benchmark (yes, 7B!!)
- 32B crushes all open-weight models tested, even those 100x larger 🤯
Open-sourcing the future of code reasoning! 🚀
Check it out https://huggingface.co/blog/open-r1/update-3
- 7B model outperforms Claude 3.7 Sonnet on IOI benchmark (yes, 7B!!)
- 32B crushes all open-weight models tested, even those 100x larger 🤯
Open-sourcing the future of code reasoning! 🚀
Check it out https://huggingface.co/blog/open-r1/update-3
BrigitteTousi
posted
an
update
5 days ago
Post
3121
LeRobot goes to driving school! 🚗🚗🚗
Hugging Face just announced a new collab with Yaak to bring the largest open-source self-driving dataset to LeRobot!
Major kudos to HF's @cadene , as well as @sandhawalia , @Shnissen and the Yaak team!
Check out the blog post here: https://huggingface.co/blog/lerobot-goes-to-driving-school
Hugging Face just announced a new collab with Yaak to bring the largest open-source self-driving dataset to LeRobot!
Major kudos to HF's @cadene , as well as @sandhawalia , @Shnissen and the Yaak team!
Check out the blog post here: https://huggingface.co/blog/lerobot-goes-to-driving-school
-
- 1 reply
Post
867
Our new Agentic leaderboard is now live!💥
If you ever asked which LLM is best for powering agents, we've just made a leaderboard that ranks them all! Built with @albertvillanova , this ranks LLMs powering a smolagents CodeAgent on subsets of various benchmarks. ✅
🏆 GPT-4.5 comes on top, even beating reasoning models like DeepSeek-R1 or o1. And Claude-3.7-Sonnet is a close second!
The leaderboard also allows you to show the scores of vanilla LLMs (without any agentic setup) on the same benchmarks: this shows the huge improvements brought by agentic setups. 💪
(Note that results will be added manually, so the leaderboard might not always have the latest LLMs)
If you ever asked which LLM is best for powering agents, we've just made a leaderboard that ranks them all! Built with @albertvillanova , this ranks LLMs powering a smolagents CodeAgent on subsets of various benchmarks. ✅
🏆 GPT-4.5 comes on top, even beating reasoning models like DeepSeek-R1 or o1. And Claude-3.7-Sonnet is a close second!
The leaderboard also allows you to show the scores of vanilla LLMs (without any agentic setup) on the same benchmarks: this shows the huge improvements brought by agentic setups. 💪
(Note that results will be added manually, so the leaderboard might not always have the latest LLMs)
- 1 reply
BrigitteTousi
posted
an
update
6 days ago
Post
3622
Regardless of X being down or not, so glad I can rely on HF Posts for AI news ❤️🤗
-
- 1 reply
Post
5673
Honored to be named among their 12 pioneers and power players in the news industry in the 2025 Tech Trends Report from Future Today Strategy Group.
Incredible group to be part of - each person is doing groundbreaking work at the intersection of AI and journalism. Worth following them all: they're consistently sharing practical insights on building the future of news.
Take the time to read this report, it's packed with insights as always. The news & information section's #1 insight hits hard: "The most substantive economic impact of AI to date has been licensing payouts for a handful of big publishers. The competition will start shifting in the year ahead to separate AI 'haves' that have positioned themselves to grow from the 'have-nots.'"
This AI-driven divide is something I've been really concerned about. Now is the time to build more than ever!
👉 Full report here: https://ftsg.com/wp-content/uploads/2025/03/FTSG_2025_TR_FINAL_LINKED.pdf
Incredible group to be part of - each person is doing groundbreaking work at the intersection of AI and journalism. Worth following them all: they're consistently sharing practical insights on building the future of news.
Take the time to read this report, it's packed with insights as always. The news & information section's #1 insight hits hard: "The most substantive economic impact of AI to date has been licensing payouts for a handful of big publishers. The competition will start shifting in the year ahead to separate AI 'haves' that have positioned themselves to grow from the 'have-nots.'"
This AI-driven divide is something I've been really concerned about. Now is the time to build more than ever!
👉 Full report here: https://ftsg.com/wp-content/uploads/2025/03/FTSG_2025_TR_FINAL_LINKED.pdf
- 2 replies
Post
4052
AI will bring us "a country of yes-men on servers" instead of one of "Einsteins sitting in a data center" if we continue on current trends.
Must-read by @thomwolf deflating overblown AI promises and explaining what real scientific breakthroughs require.
https://thomwolf.io/blog/scientific-ai.html
Must-read by @thomwolf deflating overblown AI promises and explaining what real scientific breakthroughs require.
https://thomwolf.io/blog/scientific-ai.html
- 2 replies