LFM2 Technical Report
Paper
•
2511.23404
•
Published
•
50
gg-tt
company
AI & ML interests
None defined yet.
Recent Activity
View all activity
mlabonne
authored
2
papers
3 days ago
danielhanchen
posted
an
update
4 days ago
Post
2383
You can now fine-tune embedding models in our free Unsloth notebook! 🤗
Fine-tuning embedding models improves retrieval & RAG by aligning vectors to your domain-specific notion of similarity, improving search, clustering, and recommendations on your data.
⭐ Blog + Notebooks: https://unsloth.ai/docs/new/embedding-finetuning
Unsloth trains embedding models 1.8-3.3x faster with 20% less VRAM, 2x longer context & no accuracy loss vs. FA2 setups.
We'd like to thank Hugging Face and Unsloth contributor: electroglyph for making this possible!
Fine-tuning embedding models improves retrieval & RAG by aligning vectors to your domain-specific notion of similarity, improving search, clustering, and recommendations on your data.
⭐ Blog + Notebooks: https://unsloth.ai/docs/new/embedding-finetuning
Unsloth trains embedding models 1.8-3.3x faster with 20% less VRAM, 2x longer context & no accuracy loss vs. FA2 setups.
We'd like to thank Hugging Face and Unsloth contributor: electroglyph for making this possible!
- 1 reply
danielhanchen
posted
an
update
6 days ago
Post
2496
Run GLM-4.7-Flash locally on your device with 24GB RAM!🔥
It's the best performing 30B model on SWE-Bench and GPQA. With 200K context, it excels at coding, agents, chat & reasoning.
GGUF: unsloth/GLM-4.7-Flash-GGUF
Guide: https://unsloth.ai/docs/models/glm-4.7-flash
It's the best performing 30B model on SWE-Bench and GPQA. With 200K context, it excels at coding, agents, chat & reasoning.
GGUF: unsloth/GLM-4.7-Flash-GGUF
Guide: https://unsloth.ai/docs/models/glm-4.7-flash
danielhanchen
posted
an
update
11 days ago
Post
2756
You can now do reinforcement learning training with 7× longer context and no accuracy loss, via our new batching algorithms.
Long reasoning chains in RL are costly, but now we enable you to train gpt-oss with GRPO & reach 380K context on a 192GB GPU.
Blog: https://unsloth.ai/docs/new/grpo-long-context
Long reasoning chains in RL are costly, but now we enable you to train gpt-oss with GRPO & reach 380K context on a 192GB GPU.
Blog: https://unsloth.ai/docs/new/grpo-long-context
RyanMullins
authored
a
paper
11 days ago
Post
4773
New family of 1B models just dropped!
> LiquidAI/LFM2.5-1.2B-Base: 10T → 28T tokens
> LiquidAI/LFM2.5-1.2B-Instruct: new large-scale multi-stage RL
> LiquidAI/LFM2.5-1.2B-JP: our most polite model
> LiquidAI/LFM2.5-VL-1.6B: multi-image multilingual
> LiquidAI/LFM2.5-Audio-1.5B: 8x times faster, no quality loss
Super proud of this release 🤗
> LiquidAI/LFM2.5-1.2B-Base: 10T → 28T tokens
> LiquidAI/LFM2.5-1.2B-Instruct: new large-scale multi-stage RL
> LiquidAI/LFM2.5-1.2B-JP: our most polite model
> LiquidAI/LFM2.5-VL-1.6B: multi-image multilingual
> LiquidAI/LFM2.5-Audio-1.5B: 8x times faster, no quality loss
Super proud of this release 🤗
-
- 3 replies
kathleenatgoogle
authored
a
paper
20 days ago
Post
2944
👉 What happened in AI in 2025? 👈
We prepared the 2025 version of the HF AI Timeline Grid, highlighting open vs API-based model releases, and allowing you to browse and filter by access, modality, and release type!
Play with it here:
2025-ai-timeline/2025-ai-timeline
Here's my personal quarterly TL;DR:
1️⃣ Q1 — Learning to Reason
Deepseek not only releases a top-notch reasoning model, but shows how to train them and compete with closed frontier models. OpenAI debuts Deep Research.
Significant milestones: DeepSeek R1 & R1-Zero, Qwen 2.5 VL, OpenAI Deep Research, Gemini 2.5 Pro (experimental)
2️⃣ Q2 — Multimodality and Coding
More LLMs embrace multimodality by default, and there's a surge in coding agents. Strong vision, audio, and generative models emerge.
Significant milestones: Llama 4, Qwen 3, Imagen 4, OpenAI Codex, Google Jules, Claude 4
3️⃣ Q3 — "Gold" rush, OpenAI opens up, the community goes bananas
Flagship models get gold in Math olympiads and hard benchmarks. OpenAI releases strong open source models and Google releases the much anticipated nano-banana for image generation and editing. Agentic workflows become commonplace.
Significant milestones: Gemini and OpenAI IMO Gold, gpt-oss, Gemini 2.5 Flash Image, Grok 4, Claude Sonnet 4.5
4️⃣ Q4 — Mistral returns, leaderboard hill-climbing
Mistral is back with updated model families. All labs release impressive models to wrap up the year!
Significant milestones: Claude Opus 4.5, DeepSeek Math V2, FLUX 2, GPT 5.1, Kimi K2 Thinking, Nano Banana Pro, GLM 4.7, Gemini 3, Mistral 3, MiniMax M2.1 🤯
Credits
🙏 NHLOCAL for the source data https://github.com/NHLOCAL/AiTimeline
🫡 @reach-vb for the original idea, design and recipe
🙌 @ariG23498 and yours truly for compiling and verifying the 2025 edition
🥳 Here's to 2026, wishing it becomes the best year ever for open releases and on-device-first use-cases! 🥂
We prepared the 2025 version of the HF AI Timeline Grid, highlighting open vs API-based model releases, and allowing you to browse and filter by access, modality, and release type!
Play with it here:
2025-ai-timeline/2025-ai-timeline
Here's my personal quarterly TL;DR:
1️⃣ Q1 — Learning to Reason
Deepseek not only releases a top-notch reasoning model, but shows how to train them and compete with closed frontier models. OpenAI debuts Deep Research.
Significant milestones: DeepSeek R1 & R1-Zero, Qwen 2.5 VL, OpenAI Deep Research, Gemini 2.5 Pro (experimental)
2️⃣ Q2 — Multimodality and Coding
More LLMs embrace multimodality by default, and there's a surge in coding agents. Strong vision, audio, and generative models emerge.
Significant milestones: Llama 4, Qwen 3, Imagen 4, OpenAI Codex, Google Jules, Claude 4
3️⃣ Q3 — "Gold" rush, OpenAI opens up, the community goes bananas
Flagship models get gold in Math olympiads and hard benchmarks. OpenAI releases strong open source models and Google releases the much anticipated nano-banana for image generation and editing. Agentic workflows become commonplace.
Significant milestones: Gemini and OpenAI IMO Gold, gpt-oss, Gemini 2.5 Flash Image, Grok 4, Claude Sonnet 4.5
4️⃣ Q4 — Mistral returns, leaderboard hill-climbing
Mistral is back with updated model families. All labs release impressive models to wrap up the year!
Significant milestones: Claude Opus 4.5, DeepSeek Math V2, FLUX 2, GPT 5.1, Kimi K2 Thinking, Nano Banana Pro, GLM 4.7, Gemini 3, Mistral 3, MiniMax M2.1 🤯
Credits
🙏 NHLOCAL for the source data https://github.com/NHLOCAL/AiTimeline
🫡 @reach-vb for the original idea, design and recipe
🙌 @ariG23498 and yours truly for compiling and verifying the 2025 edition
🥳 Here's to 2026, wishing it becomes the best year ever for open releases and on-device-first use-cases! 🥂
- 1 reply
danielhanchen
posted
an
update
26 days ago
Post
918
Run Qwen-Image-2512, the new SOTA text-to-image model! 💜
It's the top performing open diffusion model and has more realistic + accurate images/text.
Run locally with 14GB RAM via our Dynamic GGUF: unsloth/Qwen-Image-2512-GGUF
Guide: https://unsloth.ai/docs/models/qwen-image-2512
It's the top performing open diffusion model and has more realistic + accurate images/text.
Run locally with 14GB RAM via our Dynamic GGUF: unsloth/Qwen-Image-2512-GGUF
Guide: https://unsloth.ai/docs/models/qwen-image-2512
- 2 replies
danielhanchen
posted
an
update
about 1 month ago
Post
4060
You can now run GLM-4.7, the new 355B parameter SOTA model on your local device (128GB RAM).✨
The model achieves SOTA performance on coding, agentic and chat benchmarks.
GGUF: unsloth/GLM-4.7-GGUF
Guide: https://docs.unsloth.ai/models/glm-4.7
The model achieves SOTA performance on coding, agentic and chat benchmarks.
GGUF: unsloth/GLM-4.7-GGUF
Guide: https://docs.unsloth.ai/models/glm-4.7
- 3 replies
danielhanchen
posted
an
update
about 1 month ago
Post
2462
Google releases FunctionGemma, a new 270M parameter model that runs on just 0.5 GB RAM.✨
Built for tool-calling, run locally on your phone at 50+ tokens/s, or fine-tune with Unsloth & deploy to your phone.
GGUF: unsloth/functiongemma-270m-it-GGUF
Docs + Notebook: https://docs.unsloth.ai/models/functiongemma
Built for tool-calling, run locally on your phone at 50+ tokens/s, or fine-tune with Unsloth & deploy to your phone.
GGUF: unsloth/functiongemma-270m-it-GGUF
Docs + Notebook: https://docs.unsloth.ai/models/functiongemma
-
- 2 replies
danielhanchen
posted
an
update
about 1 month ago
Post
5504
NVIDIA releases Nemotron 3 Nano, a new 30B hybrid reasoning model! 🔥
Has 1M context window & best in class performance for SWE-Bench, reasoning & chat. Run the MoE model locally with 24GB RAM.
GGUF: unsloth/Nemotron-3-Nano-30B-A3B-GGUF
💚 Step-by-step Guide: https://docs.unsloth.ai/models/nemotron-3
Has 1M context window & best in class performance for SWE-Bench, reasoning & chat. Run the MoE model locally with 24GB RAM.
GGUF: unsloth/Nemotron-3-Nano-30B-A3B-GGUF
💚 Step-by-step Guide: https://docs.unsloth.ai/models/nemotron-3
- 1 reply
danielhanchen
posted
an
update
about 2 months ago
Post
2170
Mistral's new SOTA coding models Devstral 2 can now be Run locally! (25GB RAM) 🐱
We fixed the chat template, so performance should be much better now!
24B: unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF
123B: unsloth/Devstral-2-123B-Instruct-2512-GGUF
🧡Step-by-step Guide: https://docs.unsloth.ai/models/devstral-2
We fixed the chat template, so performance should be much better now!
24B: unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF
123B: unsloth/Devstral-2-123B-Instruct-2512-GGUF
🧡Step-by-step Guide: https://docs.unsloth.ai/models/devstral-2
Post
3343
🐦🔥 I've just published Sentence Transformers v5.2.0! It introduces multi-processing for CrossEncoder (rerankers), multilingual NanoBEIR evaluators, similarity score outputs in mine_hard_negatives, Transformers v5 support and more. Details:
- CrossEncoder multi-processing: Similar to SentenceTransformer and SparseEncoder, you can now use multi-processing with CrossEncoder rerankers. Useful for multi-GPU and CPU settings, and simple to configure: just
- Multilingual NanoBEIR Support: You can now use community translations of the tiny NanoBEIR retrieval benchmark instead of only the English one, by passing
- Similarity scores in Hard Negatives Mining: When mining for hard negatives to create a strong training dataset, you can now pass
- Transformers v5: This release works with both Transformers v4 and the upcoming v5. In the future, Sentence Transformers will only work with Transformers v5, but not yet!
- Python 3.9 deprecation: Now that Python 3.9 has lost security support, Sentence Transformers no longer supports it.
Check out the full changelog for more details: https://github.com/huggingface/sentence-transformers/releases/tag/v5.2.0
I'm quite excited about what's coming. There's a huge draft PR with a notable refactor in the works that should bring some exciting support. Specifically, better multimodality, rerankers, and perhaps some late interaction in the future!
- CrossEncoder multi-processing: Similar to SentenceTransformer and SparseEncoder, you can now use multi-processing with CrossEncoder rerankers. Useful for multi-GPU and CPU settings, and simple to configure: just
device=["cuda:0", "cuda:1"] or device=["cpu"]*4 on the model.predict or model.rank calls.- Multilingual NanoBEIR Support: You can now use community translations of the tiny NanoBEIR retrieval benchmark instead of only the English one, by passing
dataset_id, e.g. dataset_id="lightonai/NanoBEIR-de" for the German benchmark.- Similarity scores in Hard Negatives Mining: When mining for hard negatives to create a strong training dataset, you can now pass
output_scores=True to get similarity scores returned. This can be useful for some distillation losses!- Transformers v5: This release works with both Transformers v4 and the upcoming v5. In the future, Sentence Transformers will only work with Transformers v5, but not yet!
- Python 3.9 deprecation: Now that Python 3.9 has lost security support, Sentence Transformers no longer supports it.
Check out the full changelog for more details: https://github.com/huggingface/sentence-transformers/releases/tag/v5.2.0
I'm quite excited about what's coming. There's a huge draft PR with a notable refactor in the works that should bring some exciting support. Specifically, better multimodality, rerankers, and perhaps some late interaction in the future!
danielhanchen
posted
an
update
about 2 months ago
Post
3839
Mistral's new Ministral 3 models can now be Run & Fine-tuned locally! (16GB RAM)
Ministral 3 have vision support and the best-in-class performance for their sizes.
14B Instruct GGUF: unsloth/Ministral-3-14B-Instruct-2512-GGUF
14B Reasoning GGUF: unsloth/Ministral-3-14B-Reasoning-2512-GGUF
🐱 Step-by-step Guide: https://docs.unsloth.ai/new/ministral-3
All GGUFs, BnB, FP8 etc. variants uploads: https://huggingface.co/collections/unsloth/ministral-3
Ministral 3 have vision support and the best-in-class performance for their sizes.
14B Instruct GGUF: unsloth/Ministral-3-14B-Instruct-2512-GGUF
14B Reasoning GGUF: unsloth/Ministral-3-14B-Reasoning-2512-GGUF
🐱 Step-by-step Guide: https://docs.unsloth.ai/new/ministral-3
All GGUFs, BnB, FP8 etc. variants uploads: https://huggingface.co/collections/unsloth/ministral-3
-
- 3 replies
danielhanchen
posted
an
update
about 2 months ago
Post
8568
Qwen3-Next can now be Run locally! (30GB RAM)
Instruct GGUF: unsloth/Qwen3-Next-80B-A3B-Instruct-GGUF
The models come in Thinking and Instruct versions and utilize a new architecture, allowing it to have ~10x faster inference than Qwen32B.
💜 Step-by-step Guide: https://docs.unsloth.ai/models/qwen3-next
Thinking GGUF: unsloth/Qwen3-Next-80B-A3B-Thinking-GGUF
Instruct GGUF: unsloth/Qwen3-Next-80B-A3B-Instruct-GGUF
The models come in Thinking and Instruct versions and utilize a new architecture, allowing it to have ~10x faster inference than Qwen32B.
💜 Step-by-step Guide: https://docs.unsloth.ai/models/qwen3-next
Thinking GGUF: unsloth/Qwen3-Next-80B-A3B-Thinking-GGUF
danielhanchen
posted
an
update
3 months ago
Post
4445
You can now run Kimi K2 Thinking locally with our Dynamic 1-bit GGUFs:
unsloth/Kimi-K2-Thinking-GGUF
We shrank the 1T model to 245GB (-62%) & retained ~85% of accuracy on Aider Polyglot. Run on >247GB RAM for fast inference.
We also collaborated with the Moonshot AI Kimi team on a system prompt fix! 🥰
Guide + fix details: https://docs.unsloth.ai/models/kimi-k2-thinking-how-to-run-locally
We shrank the 1T model to 245GB (-62%) & retained ~85% of accuracy on Aider Polyglot. Run on >247GB RAM for fast inference.
We also collaborated with the Moonshot AI Kimi team on a system prompt fix! 🥰
Guide + fix details: https://docs.unsloth.ai/models/kimi-k2-thinking-how-to-run-locally
Post
4411
🤗 Sentence Transformers is joining Hugging Face! 🤗 This formalizes the existing maintenance structure, as I've personally led the project for the past two years on behalf of Hugging Face! Details:
Today, the Ubiquitous Knowledge Processing (UKP) Lab is transferring the project to Hugging Face. Sentence Transformers will remain a community-driven, open-source project, with the same open-source license (Apache 2.0) as before. Contributions from researchers, developers, and enthusiasts are welcome and encouraged. The project will continue to prioritize transparency, collaboration, and broad accessibility.
Read our full announcement for more details and quotes from UKP and Hugging Face leadership: https://huggingface.co/blog/sentence-transformers-joins-hf
We see an increasing wish from companies to move from large LLM APIs to local models for better control and privacy, reflected in the library's growth: in just the last 30 days, Sentence Transformer models have been downloaded >270 million times, second only to transformers.
I would like to thank the UKP Lab, and especially Nils Reimers and Iryna Gurevych, both for their dedication to the project and for their trust in myself, both now and two years ago. Back then, neither of you knew me well, yet you trusted me to take the project to new heights. That choice ended up being very valuable for the embedding & Information Retrieval community, and I think this choice of granting Hugging Face stewardship will be similarly successful.
I'm very excited about the future of the project, and for the world of embeddings and retrieval at large!
Today, the Ubiquitous Knowledge Processing (UKP) Lab is transferring the project to Hugging Face. Sentence Transformers will remain a community-driven, open-source project, with the same open-source license (Apache 2.0) as before. Contributions from researchers, developers, and enthusiasts are welcome and encouraged. The project will continue to prioritize transparency, collaboration, and broad accessibility.
Read our full announcement for more details and quotes from UKP and Hugging Face leadership: https://huggingface.co/blog/sentence-transformers-joins-hf
We see an increasing wish from companies to move from large LLM APIs to local models for better control and privacy, reflected in the library's growth: in just the last 30 days, Sentence Transformer models have been downloaded >270 million times, second only to transformers.
I would like to thank the UKP Lab, and especially Nils Reimers and Iryna Gurevych, both for their dedication to the project and for their trust in myself, both now and two years ago. Back then, neither of you knew me well, yet you trusted me to take the project to new heights. That choice ended up being very valuable for the embedding & Information Retrieval community, and I think this choice of granting Hugging Face stewardship will be similarly successful.
I'm very excited about the future of the project, and for the world of embeddings and retrieval at large!
- 1 reply
Post
8306
LiquidAI/LFM2-8B-A1B just dropped!
8.3B params with only 1.5B active/token 🚀
> Quality ≈ 3–4B dense, yet faster than Qwen3-1.7B
> MoE designed to run on phones/laptops (llama.cpp / vLLM)
> Pre-trained on 12T tokens → strong math/code/IF
8.3B params with only 1.5B active/token 🚀
> Quality ≈ 3–4B dense, yet faster than Qwen3-1.7B
> MoE designed to run on phones/laptops (llama.cpp / vLLM)
> Pre-trained on 12T tokens → strong math/code/IF
- 1 reply