One Year Since the “DeepSeek Moment”
•
8
AI & ML interests
The AI community building the future.
Recent Activity
View all activity
Papers
View all Papers
Articles
Post
DeepSeek R1 dropped one year ago 🐳 and a lot has changed.
With @irenesolaiman , we’re launching a blog series about how that moment reshaped AI + open source in 2025, starting with strategic shifts and the explosion of new open models in China!
https://huggingface.co/blog/huggingface/one-year-since-the-deepseek-moment
With @irenesolaiman , we’re launching a blog series about how that moment reshaped AI + open source in 2025, starting with strategic shifts and the explosion of new open models in China!
https://huggingface.co/blog/huggingface/one-year-since-the-deepseek-moment
Article
One Year Since the “DeepSeek Moment”
•
8
burtenshaw
updated
a
model
about 3 hours ago
sayakpaul
updated
a
dataset
about 6 hours ago
Post
1347
Z.ai just released a powerful lightweight option of GLM 4.7
✨ 30B total/3B active - MoE
zai-org/GLM-4.7-Flash
✨ 30B total/3B active - MoE
zai-org/GLM-4.7-Flash
sergiopaniego
posted
an
update
1 day ago
Post
534
FunctionGemma Tuning Lab is a new no-code tool by
@google
that lets you fine-tune a model directly from the browser, with no coding knowledge required, using TRL behind the scenes.
blog: https://developers.googleblog.com/a-guide-to-fine-tuning-functiongemma/
try it out: google/functiongemma-tuning-lab
This example builds on a more advanced one for learning fine-tuning with SFT using TRL: https://ai.google.dev/gemma/docs/functiongemma/finetuning-with-functiongemma
blog: https://developers.googleblog.com/a-guide-to-fine-tuning-functiongemma/
try it out: google/functiongemma-tuning-lab
This example builds on a more advanced one for learning fine-tuning with SFT using TRL: https://ai.google.dev/gemma/docs/functiongemma/finetuning-with-functiongemma
- 1 reply
Post
115
Another Chinese model fully trained on domestic chips, released by China Telecom 👀
Tele-AI/TeleChat3-36B-Thinking
TeleChat3-36B-Thinking:
✨ Native support for the Ascend + MindSpore ecosystem
✨ Inspired by DeepSeek’s architecture design, bringing training stability and efficiency gains.
Tele-AI/TeleChat3-36B-Thinking
TeleChat3-36B-Thinking:
✨ Native support for the Ascend + MindSpore ecosystem
✨ Inspired by DeepSeek’s architecture design, bringing training stability and efficiency gains.
-
- 2 replies
sergiopaniego
posted
an
update
4 days ago
Post
623
TRL v0.27.0 is out!! 🥳
It includes GDPO, the latest variant of GRPO for multi-reward RL ✨
GDPO decouples reward normalization to avoid reward collapse and improve per-reward convergence — developed by
@sliuau @SimonX et al.
Explore the paper: GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization (2601.05242)
Explore the full set of changes here:
https://github.com/huggingface/trl/releases/tag/v0.27.0
It includes GDPO, the latest variant of GRPO for multi-reward RL ✨
GDPO decouples reward normalization to avoid reward collapse and improve per-reward convergence — developed by
@sliuau @SimonX et al.
Explore the paper: GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization (2601.05242)
Explore the full set of changes here:
https://github.com/huggingface/trl/releases/tag/v0.27.0
Post
974
After a VLM, StepFun dropped a new audio model: Step-Audio-R1.1, enabling thinking while speaking 🔥
stepfun-ai/Step-Audio-R1.1
✨ Apache 2.0
✨ Combines dual-brain architecture and acoustic-grounded reasoning to enable real-time dialogue with SOTA-level reasoning
stepfun-ai/Step-Audio-R1.1
✨ Apache 2.0
✨ Combines dual-brain architecture and acoustic-grounded reasoning to enable real-time dialogue with SOTA-level reasoning
-
- 2 replies
akhaliq
submitted
a
paper
to Daily Papers
4 days ago
Post
1680
We have a new heatmap live on huggingface now🔥
woojun-jung/open-source-release-heatmap-ko
Korean community built their own version to track labs that actively publish open work, inspired by Chinese open source heat map!
This is the open source community at its best ♥️
woojun-jung/open-source-release-heatmap-ko
Korean community built their own version to track labs that actively publish open work, inspired by Chinese open source heat map!
This is the open source community at its best ♥️
- 1 reply
Post
655
More lightweight multimodal models are coming 👀
StepFun has been focused on multimodal AI from the very beginning. Their latest release a new foundational model: STEP3-VL🔥
https://huggingface.co/collections/stepfun-ai/step3-vl-10b
✨ 10B - Apache2.0
✨ Leads in the 10B class and competes with models 10–20× larger
StepFun has been focused on multimodal AI from the very beginning. Their latest release a new foundational model: STEP3-VL🔥
https://huggingface.co/collections/stepfun-ai/step3-vl-10b
✨ 10B - Apache2.0
✨ Leads in the 10B class and competes with models 10–20× larger
Post
307
Agentic capability is the new battleground🔥
LongCat-Flash-Thinking-2601, the latest reasoning model from Meituan- LongCat
✨ MoE - 560B total / 27B active
✨ MIT license
✨ Agentic tool use
✨ Multi-environment RL
✨ Parallel + iterative reasoning
meituan-longcat/LongCat-Flash-Thinking-2601
LongCat-Flash-Thinking-2601, the latest reasoning model from Meituan- LongCat
✨ MoE - 560B total / 27B active
✨ MIT license
✨ Agentic tool use
✨ Multi-environment RL
✨ Parallel + iterative reasoning
meituan-longcat/LongCat-Flash-Thinking-2601
akhaliq
submitted
a
paper
to Daily Papers
6 days ago
Post
305
GLM-Image from Z.ai is out 🔥
It was fully trained on Ascend Atlas 800T A2 with MindSpore, probably the first SOTA multimodal model fully trained on domestic chips 👀
zai-org/GLM-Image
✨ Hybrid Architecture: combined autoregressive + diffusion design delivers strong semantic alignment with high-fidelity details
✨ Strong performance in long, dense, and multilingual text rendering
✨ MIT licensed (VQ tokenizer & ViT weights under Apache 2.0)
✨ Now live on Hugging Face inference provider 🤗
It was fully trained on Ascend Atlas 800T A2 with MindSpore, probably the first SOTA multimodal model fully trained on domestic chips 👀
zai-org/GLM-Image
✨ Hybrid Architecture: combined autoregressive + diffusion design delivers strong semantic alignment with high-fidelity details
✨ Strong performance in long, dense, and multilingual text rendering
✨ MIT licensed (VQ tokenizer & ViT weights under Apache 2.0)
✨ Now live on Hugging Face inference provider 🤗
sergiopaniego
posted
an
update
7 days ago
Post
2876
New REPL environment in OpenEnv available! ✨
Used in the Recursive Language Models (RLM) paper by Alex Zhang.
Ready for inference & post-training using trajectories. Handles long contexts:
> Run Python code in a sandbox
> Make recursive calls to LMs
> Explore data programmatically
> Return final result
Docs: https://meta-pytorch.org/OpenEnv/environments/repl/
Inference script: https://github.com/meta-pytorch/OpenEnv/blob/main/examples/repl_oolong_simple.py
Used in the Recursive Language Models (RLM) paper by Alex Zhang.
Ready for inference & post-training using trajectories. Handles long contexts:
> Run Python code in a sandbox
> Make recursive calls to LMs
> Explore data programmatically
> Return final result
Docs: https://meta-pytorch.org/OpenEnv/environments/repl/
Inference script: https://github.com/meta-pytorch/OpenEnv/blob/main/examples/repl_oolong_simple.py
Post
2619
From ChatGPT Healthcare to Claude for healthcare, AI in medicine is speeding up🚀
Now BaichuanAI joins with Baichuan-M3 🏥 an open medical LLM trained for clinical decision-making
https://huggingface.co/collections/baichuan-inc/baichuan-m3
✨ 235B - Apache2.0
✨ Lower hallucinations via Fact-Aware RL
✨ Built for long medical chats
Now BaichuanAI joins with Baichuan-M3 🏥 an open medical LLM trained for clinical decision-making
https://huggingface.co/collections/baichuan-inc/baichuan-m3
✨ 235B - Apache2.0
✨ Lower hallucinations via Fact-Aware RL
✨ Built for long medical chats
- 2 replies
sergiopaniego
posted
an
update
8 days ago
Post
361
Recursive Language Models (RLM) is a new interface for LLMs with cool ideas by Alex Zhang!
⚠️ LLMs struggle with long prompts → attention overload & lost info
🔄 RLMs inspect, split & call themselves on chunks, then aggregate results
✅ Handles millions of tokens, reduces noise, improves reasoning
💡 System prompt guides recursion
🎯 RLM trajectories can be used for RL training or distillation (OpenEnv+TRL!!)
We're adding it to OpenEnv (with Kashif Rasul): https://github.com/meta-pytorch/OpenEnv/pull/282
More resources:
> Paper: Recursive Language Models (2512.24601)
> Paper blog: https://alexzhang13.github.io/blog/2025/rlm/
> RLM repo: https://github.com/alexzhang13/rlm
⚠️ LLMs struggle with long prompts → attention overload & lost info
🔄 RLMs inspect, split & call themselves on chunks, then aggregate results
✅ Handles millions of tokens, reduces noise, improves reasoning
💡 System prompt guides recursion
🎯 RLM trajectories can be used for RL training or distillation (OpenEnv+TRL!!)
We're adding it to OpenEnv (with Kashif Rasul): https://github.com/meta-pytorch/OpenEnv/pull/282
More resources:
> Paper: Recursive Language Models (2512.24601)
> Paper blog: https://alexzhang13.github.io/blog/2025/rlm/
> RLM repo: https://github.com/alexzhang13/rlm
-
- 2 replies
Post
2810
AgentCPM-Explore🔥 on device agent foundation model released by OpenBMB
openbmb/AgentCPM-Explore
✨ 4B - Apache2.0
✨ Supports 100+ multi-turn environment interactions with search + verification
✨ Full training/inference stack is openly shared as well
openbmb/AgentCPM-Explore
✨ 4B - Apache2.0
✨ Supports 100+ multi-turn environment interactions with search + verification
✨ Full training/inference stack is openly shared as well
Post
2570
Based on 2025 Chinese AI Timeline, here are some interesting takeaways:
✨ DeepSeek cadence: They shipped almost every month! (except Feb 2025)
✨ Qwen trajectory: Not a single “hit” model, but an expanding product line. VL/Math/Coder/Reranker/Embedding/Omni/Next/Image
✨ Multimodal trend: Steadily rising share, shifting from generation to editing + tooling.
✨ Reasoning as a main track: more engineered, system-level reasoning.
✨ From foundation to components: growth in infra models (embeddings, rerankers, OCR, speech) signals a move toward deployable stacks.
✨ Ecosystem broadening: more players beyond the top labs.
Follow for more updates👉
zh-ai-community
✨ DeepSeek cadence: They shipped almost every month! (except Feb 2025)
✨ Qwen trajectory: Not a single “hit” model, but an expanding product line. VL/Math/Coder/Reranker/Embedding/Omni/Next/Image
✨ Multimodal trend: Steadily rising share, shifting from generation to editing + tooling.
✨ Reasoning as a main track: more engineered, system-level reasoning.
✨ From foundation to components: growth in infra models (embeddings, rerankers, OCR, speech) signals a move toward deployable stacks.
✨ Ecosystem broadening: more players beyond the top labs.
Follow for more updates👉
-
- 2 replies