Optimizing Test-Time Compute via Meta Reinforcement Fine-Tuning
Paper
•
2503.07572
•
Published
•
25
fast.ai community
non-profit
AI & ML interests
Making neural nets uncool again.
Recent Activity
View all activity
fastai's activity
lewtun
authored
a
paper
about 14 hours ago
Post
1415
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
4857
Introducing OpenR1-Math-220k!
open-r1/OpenR1-Math-220k
The community has been busy distilling DeepSeek-R1 from inference providers, but we decided to have a go at doing it ourselves from scratch 💪
What’s new compared to existing reasoning datasets?
♾ Based on AI-MO/NuminaMath-1.5: we focus on math reasoning traces and generate answers for problems in NuminaMath 1.5, an improved version of the popular NuminaMath-CoT dataset.
🐳 800k R1 reasoning traces: We generate two answers for 400k problems using DeepSeek R1. The filtered dataset contains 220k problems with correct reasoning traces.
📀 512 H100s running locally: Instead of relying on an API, we leverage vLLM and SGLang to run generations locally on our science cluster, generating 180k reasoning traces per day.
⏳ Automated filtering: We apply Math Verify to only retain problems with at least one correct answer. We also leverage Llama3.3-70B-Instruct as a judge to retrieve more correct examples (e.g for cases with malformed answers that can’t be verified with a rules-based parser)
📊 We match the performance of DeepSeek-Distill-Qwen-7B by finetuning Qwen-7B-Math-Instruct on our dataset.
🔎 Read our blog post for all the nitty gritty details: https://huggingface.co./blog/open-r1/update-2
open-r1/OpenR1-Math-220k
The community has been busy distilling DeepSeek-R1 from inference providers, but we decided to have a go at doing it ourselves from scratch 💪
What’s new compared to existing reasoning datasets?
♾ Based on AI-MO/NuminaMath-1.5: we focus on math reasoning traces and generate answers for problems in NuminaMath 1.5, an improved version of the popular NuminaMath-CoT dataset.
🐳 800k R1 reasoning traces: We generate two answers for 400k problems using DeepSeek R1. The filtered dataset contains 220k problems with correct reasoning traces.
📀 512 H100s running locally: Instead of relying on an API, we leverage vLLM and SGLang to run generations locally on our science cluster, generating 180k reasoning traces per day.
⏳ Automated filtering: We apply Math Verify to only retain problems with at least one correct answer. We also leverage Llama3.3-70B-Instruct as a judge to retrieve more correct examples (e.g for cases with malformed answers that can’t be verified with a rules-based parser)
📊 We match the performance of DeepSeek-Distill-Qwen-7B by finetuning Qwen-7B-Math-Instruct on our dataset.
🔎 Read our blog post for all the nitty gritty details: https://huggingface.co./blog/open-r1/update-2
lewtun
authored
a
paper
about 1 month ago
ameerazam08
posted
an
update
about 1 month ago
Post
10238
We are reproducing the full DeepSeek R1 data and training pipeline so everybody can use their recipe. Instead of doing it in secret we can do it together in the open!
🧪 Step 1: replicate the R1-Distill models by distilling a high-quality reasoning corpus from DeepSeek-R1.
🧠 Step 2: replicate the pure RL pipeline that DeepSeek used to create R1-Zero. This will involve curating new, large-scale datasets for math, reasoning, and code.
🔥 Step 3: show we can go from base model -> SFT -> RL via multi-stage training.
Follow along: https://github.com/huggingface/open-r1
🧪 Step 1: replicate the R1-Distill models by distilling a high-quality reasoning corpus from DeepSeek-R1.
🧠 Step 2: replicate the pure RL pipeline that DeepSeek used to create R1-Zero. This will involve curating new, large-scale datasets for math, reasoning, and code.
🔥 Step 3: show we can go from base model -> SFT -> RL via multi-stage training.
Follow along: https://github.com/huggingface/open-r1
- 5 replies
Post
3888
I was initially pretty sceptical about Meta's Coconut paper [1] because the largest perf gains were reported on toy linguistic problems. However, these results on machine translation are pretty impressive!
https://x.com/casper_hansen_/status/1875872309996855343
Together with the recent PRIME method [2] for scaling RL, reasoning for open models is looking pretty exciting for 2025!
[1] Training Large Language Models to Reason in a Continuous Latent Space (2412.06769)
[2] https://huggingface.co./blog/ganqu/prime
https://x.com/casper_hansen_/status/1875872309996855343
Together with the recent PRIME method [2] for scaling RL, reasoning for open models is looking pretty exciting for 2025!
[1] Training Large Language Models to Reason in a Continuous Latent Space (2412.06769)
[2] https://huggingface.co./blog/ganqu/prime
Post
2310
This paper (
HuatuoGPT-o1, Towards Medical Complex Reasoning with LLMs (2412.18925)) has a really interesting recipe for inducing o1-like behaviour in Llama models:
* Iteratively sample CoTs from the model, using a mix of different search strategies. This gives you something like Stream of Search via prompting.
* Verify correctness of each CoT using GPT-4o (needed because exact match doesn't work well in medicine where there are lots of aliases)
* Use GPT-4o to reformat the concatenated CoTs into a single stream that includes smooth transitions like "hmm, wait" etc that one sees in o1
* Use the resulting data for SFT & RL
* Use sparse rewards from GPT-4o to guide RL training. They find RL gives an average ~3 point boost across medical benchmarks and SFT on this data already gives a strong improvement.
Applying this strategy to other domains could be quite promising, provided the training data can be formulated with verifiable problems!
* Iteratively sample CoTs from the model, using a mix of different search strategies. This gives you something like Stream of Search via prompting.
* Verify correctness of each CoT using GPT-4o (needed because exact match doesn't work well in medicine where there are lots of aliases)
* Use GPT-4o to reformat the concatenated CoTs into a single stream that includes smooth transitions like "hmm, wait" etc that one sees in o1
* Use the resulting data for SFT & RL
* Use sparse rewards from GPT-4o to guide RL training. They find RL gives an average ~3 point boost across medical benchmarks and SFT on this data already gives a strong improvement.
Applying this strategy to other domains could be quite promising, provided the training data can be formulated with verifiable problems!
- 1 reply
Post
6930
We outperform Llama 70B with Llama 3B on hard math by scaling test-time compute 🔥
How? By combining step-wise reward models with tree search algorithms :)
We show that smol models can match or exceed the performance of their much larger siblings when given enough "time to think"
We're open sourcing the full recipe and sharing a detailed blog post.
In our blog post we cover:
📈 Compute-optimal scaling: How we implemented DeepMind's recipe to boost the mathematical capabilities of open models at test-time.
🎄 Diverse Verifier Tree Search (DVTS): An unpublished extension we developed to the verifier-guided tree search technique. This simple yet effective method improves diversity and delivers better performance, particularly at large test-time compute budgets.
🧭 Search and Learn: A lightweight toolkit for implementing search strategies with LLMs and built for speed with vLLM
Here's the links:
- Blog post: HuggingFaceH4/blogpost-scaling-test-time-compute
- Code: https://github.com/huggingface/search-and-learn
Enjoy!
How? By combining step-wise reward models with tree search algorithms :)
We show that smol models can match or exceed the performance of their much larger siblings when given enough "time to think"
We're open sourcing the full recipe and sharing a detailed blog post.
In our blog post we cover:
📈 Compute-optimal scaling: How we implemented DeepMind's recipe to boost the mathematical capabilities of open models at test-time.
🎄 Diverse Verifier Tree Search (DVTS): An unpublished extension we developed to the verifier-guided tree search technique. This simple yet effective method improves diversity and delivers better performance, particularly at large test-time compute budgets.
🧭 Search and Learn: A lightweight toolkit for implementing search strategies with LLMs and built for speed with vLLM
Here's the links:
- Blog post: HuggingFaceH4/blogpost-scaling-test-time-compute
- Code: https://github.com/huggingface/search-and-learn
Enjoy!
- 2 replies
muellerzr
authored
a
paper
4 months ago
ameerazam08
posted
an
update
11 months ago
Post
5043
Explore the Latest Top Papers with Papers Leaderboard!
We are excited to introduce a new way to explore the most impactful research papers: Papers Leaderboard! This feature allows you to easily find the most talked-about papers across a variety of fields.
Hf-demo : ameerazam08/Paper-LeaderBoard
Happy weekends!
We are excited to introduce a new way to explore the most impactful research papers: Papers Leaderboard! This feature allows you to easily find the most talked-about papers across a variety of fields.
Hf-demo : ameerazam08/Paper-LeaderBoard
Happy weekends!
osanseviero
posted
an
update
11 months ago
Post
12203
Diaries of Open Source. Part 15 🤗
🕵️♀️Idefics 2 is out, a multimodal open-source model with very nice capabilities
Models, demo, and datasets: HuggingFaceM4/idefics2-661d1971b7c50831dd3ce0fe
Blog: https://hf.co/blog/idefics2
💾Snowflake released snowflake-arctic-embed, a family of powerful small embedding models
Model: Snowflake/snowflake-arctic-embed-m
Blog: https://www.snowflake.com/blog/introducing-snowflake-arctic-embed-snowflakes-state-of-the-art-text-embedding-family-of-models/
✨Pile-T5, EleutherAI's T5 model trained on 2T tokens
Blog: https://blog.eleuther.ai/pile-t5/
Models: EleutherAI/pile-t5-65a76a0d0022dd270b385a66
GitHub: https://github.com/EleutherAI/improved-t5
🤖CodeQwen1.5-7B base and chat models. Models trained on 3T tokens strong benchmark results for code generation, editing and SQL
Blog post: https://qwenlm.github.io/blog/codeqwen1.5/
Demo: Qwen/CodeQwen1.5-7b-Chat-demo
Models: Qwen/CodeQwen1.5-7B and Qwen/CodeQwen1.5-7B-Chat
Misc
🦉 DocOwl1.5: Unified Stucture Learning for OCR-free Document Understanding mPLUG/DocOwl
👀Cerule - a tiny Vision LM model Tensoic/Cerule-v0.1
ChemLLM - a LLM for chemistry and molecule science ⚗️https://hf.co/AI4Chem/ChemLLM-7B-Chat-1.5-DPO
Distil Whisper Large
📝New pdf/OCR datasets with 19 samples pixparse/pdf-document-ocr-datasets-660701430b0346f97c4bc628
🔥Gretel AI high quality text-to-sql synthetic dataset gretelai/synthetic_text_to_sql
🕵️♀️Idefics 2 is out, a multimodal open-source model with very nice capabilities
Models, demo, and datasets: HuggingFaceM4/idefics2-661d1971b7c50831dd3ce0fe
Blog: https://hf.co/blog/idefics2
💾Snowflake released snowflake-arctic-embed, a family of powerful small embedding models
Model: Snowflake/snowflake-arctic-embed-m
Blog: https://www.snowflake.com/blog/introducing-snowflake-arctic-embed-snowflakes-state-of-the-art-text-embedding-family-of-models/
✨Pile-T5, EleutherAI's T5 model trained on 2T tokens
Blog: https://blog.eleuther.ai/pile-t5/
Models: EleutherAI/pile-t5-65a76a0d0022dd270b385a66
GitHub: https://github.com/EleutherAI/improved-t5
🤖CodeQwen1.5-7B base and chat models. Models trained on 3T tokens strong benchmark results for code generation, editing and SQL
Blog post: https://qwenlm.github.io/blog/codeqwen1.5/
Demo: Qwen/CodeQwen1.5-7b-Chat-demo
Models: Qwen/CodeQwen1.5-7B and Qwen/CodeQwen1.5-7B-Chat
Misc
🦉 DocOwl1.5: Unified Stucture Learning for OCR-free Document Understanding mPLUG/DocOwl
👀Cerule - a tiny Vision LM model Tensoic/Cerule-v0.1
ChemLLM - a LLM for chemistry and molecule science ⚗️https://hf.co/AI4Chem/ChemLLM-7B-Chat-1.5-DPO
Distil Whisper Large
📝New pdf/OCR datasets with 19 samples pixparse/pdf-document-ocr-datasets-660701430b0346f97c4bc628
🔥Gretel AI high quality text-to-sql synthetic dataset gretelai/synthetic_text_to_sql
-
- 5 replies
Post
5094
Introducing Zephyr 141B-A35B 🪁:
HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1
Yesterday, Mistral released their latest base model (via magnet link of course 😅) and the community quickly converted it to transformers format and pushed it to the Hub: mistral-community/Mixtral-8x22B-v0.1
Early evals of this model looked extremely strong, so we teamed up with Argilla and KAIST AI to cook up a Zephyr recipe with a few new alignment techniques that came out recently:
🧑🍳 Align the base model with Odds Ratio Preference Optimisation (ORPO). This novel algorithm developed by @JW17 and @nlee-208 and @j6mes and does not require an SFT step to achieve high performance and is thus much more computationally efficient than methods like DPO and PPO.
🦫 Use a brand new dataset of 7k high-quality, multi-turn preferences that has been developed by our friends at Argilla. To create this dataset, they took the excellent Capybara SFT dataset from @LDJnr LDJnr/Capybara and converted it into a preference dataset by augmenting the final turn with responses from new LLMs that were then ranked by GPT-4.
What we find especially neat about this approach is that training on 7k samples only takes ~1.3h on 4 H100 nodes, yet produces a model that is very strong on chat benchmarks like IFEval and BBH.
Kudos to @alvarobartt @JW17 and @nlee-208 for this very nice and fast-paced collab!
For more details on the paper and dataset, checkout our collection: HuggingFaceH4/zephyr-orpo-6617eba2c5c0e2cc3c151524
HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1
Yesterday, Mistral released their latest base model (via magnet link of course 😅) and the community quickly converted it to transformers format and pushed it to the Hub: mistral-community/Mixtral-8x22B-v0.1
Early evals of this model looked extremely strong, so we teamed up with Argilla and KAIST AI to cook up a Zephyr recipe with a few new alignment techniques that came out recently:
🧑🍳 Align the base model with Odds Ratio Preference Optimisation (ORPO). This novel algorithm developed by @JW17 and @nlee-208 and @j6mes and does not require an SFT step to achieve high performance and is thus much more computationally efficient than methods like DPO and PPO.
🦫 Use a brand new dataset of 7k high-quality, multi-turn preferences that has been developed by our friends at Argilla. To create this dataset, they took the excellent Capybara SFT dataset from @LDJnr LDJnr/Capybara and converted it into a preference dataset by augmenting the final turn with responses from new LLMs that were then ranked by GPT-4.
What we find especially neat about this approach is that training on 7k samples only takes ~1.3h on 4 H100 nodes, yet produces a model that is very strong on chat benchmarks like IFEval and BBH.
Kudos to @alvarobartt @JW17 and @nlee-208 for this very nice and fast-paced collab!
For more details on the paper and dataset, checkout our collection: HuggingFaceH4/zephyr-orpo-6617eba2c5c0e2cc3c151524
osanseviero
posted
an
update
11 months ago
Post
11249
Diaries of Open Source. Part 14 🤗
🔥CohereForAI releases Command R+, an open 104B model with:
- Tool usage capabilities
- Specialized in RAGs
- Multilingual
It's one of the first models to surpass GPT-4 in the lmsys arena, check it out!
Model: CohereForAI/c4ai-command-r-plus
Official demo: https://hf.co/spaces/CohereForAI/c4ai-command-r-plus
Quantized: CohereForAI/c4ai-command-r-plus-4bit
🎉Google releases a new version of their Gemma instruct models, with improved quality, nicer to converse, and a fancier RL algorithm. The model is similar to Llama 2 70B in the Chat Arena!
Models: google/gemma-release-65d5efbccdbb8c4202ec078b
Try it out in HuggingChat https://hf.co/chat/models/google/gemma-1.1-7b-it
🪄VoiceCraft, a speech editing and TTS SOTA open model
Paper: VoiceCraft: Zero-Shot Speech Editing and Text-to-Speech in the Wild (2403.16973)
Model: pyp1/VoiceCraft
💻Google released CodeGemma, a family of code generation, completion, and chat models
Blog post: https://hf.co/blog/codegemma
Models: google/codegemma-release-66152ac7b683e2667abdee11
Report: https://storage.googleapis.com/deepmind-media/gemma/codegemma_report.pdf
Misc models:
🦖T-Rex2, a very powerful object detection model for many applications https://github.com/IDEA-Research/T-Rex
👀 CT-RATE : A 3D dataset paired with text reports ibrahimhamamci/CT-RATE
🐙Octopus v2: a Gemma-based model trained for Android API - extremely fast, better than Llama+RAG, great results NexaAIDev/Octopus-v2
🔥CohereForAI releases Command R+, an open 104B model with:
- Tool usage capabilities
- Specialized in RAGs
- Multilingual
It's one of the first models to surpass GPT-4 in the lmsys arena, check it out!
Model: CohereForAI/c4ai-command-r-plus
Official demo: https://hf.co/spaces/CohereForAI/c4ai-command-r-plus
Quantized: CohereForAI/c4ai-command-r-plus-4bit
🎉Google releases a new version of their Gemma instruct models, with improved quality, nicer to converse, and a fancier RL algorithm. The model is similar to Llama 2 70B in the Chat Arena!
Models: google/gemma-release-65d5efbccdbb8c4202ec078b
Try it out in HuggingChat https://hf.co/chat/models/google/gemma-1.1-7b-it
🪄VoiceCraft, a speech editing and TTS SOTA open model
Paper: VoiceCraft: Zero-Shot Speech Editing and Text-to-Speech in the Wild (2403.16973)
Model: pyp1/VoiceCraft
💻Google released CodeGemma, a family of code generation, completion, and chat models
Blog post: https://hf.co/blog/codegemma
Models: google/codegemma-release-66152ac7b683e2667abdee11
Report: https://storage.googleapis.com/deepmind-media/gemma/codegemma_report.pdf
Misc models:
🦖T-Rex2, a very powerful object detection model for many applications https://github.com/IDEA-Research/T-Rex
👀 CT-RATE : A 3D dataset paired with text reports ibrahimhamamci/CT-RATE
🐙Octopus v2: a Gemma-based model trained for Android API - extremely fast, better than Llama+RAG, great results NexaAIDev/Octopus-v2
-
- 2 replies
osanseviero
posted
an
update
12 months ago
Post
2301
Diaries of Open Source. Part 13 🤗
🤏Two different bitnet 1.5 open-source replications
Original paper: The Era of 1-bit LLMs: All Large Language Models are in 1.58 Bits (2402.17764)
1bitllm experiment: https://hf.co/blog/joey00072/experiments-with-bitnet-1-5
NousResearch experiment NousResearch/OLMo-Bitnet-1B
🥳Tiny and large multimodal models great for embeddings
GitHub: https://github.com/unum-cloud/uform
Encoders: https://hf.co/collections/unum-cloud/multimodal-encoders-660553903617c5297eb16838
ONNX weights: https://hf.co/collections/unum-cloud/uform-vl-english-large-onnx-66055a57c182d846f3bc1949
📜 SMPLer-X: Expressive Human Pose and Shape Estimation
Project website: https://caizhongang.com/projects/SMPLer-X/
Demo: caizhongang/SMPLer-X
Paper: SMPLer-X: Scaling Up Expressive Human Pose and Shape Estimation (2309.17448)
🧙GeoWizard: 3D Geometry Estimation
Project website: https://fuxiao0719.github.io/projects/geowizard/
Demo: lemonaddie/geowizard
Misc models and datasets
- Dolphin-2.8-mistral-7b-v0.2 cognitivecomputations/dolphin-2.8-mistral-7b-v02
- Hermes-2-Pro-11B, a self-frankenmerge 11B variant mattshumer/Hermes-2-Pro-11B
- Large conversational dataset based on Usenet data in the Italian language mii-community/UsenetArchiveIT-conversations
🤏Two different bitnet 1.5 open-source replications
Original paper: The Era of 1-bit LLMs: All Large Language Models are in 1.58 Bits (2402.17764)
1bitllm experiment: https://hf.co/blog/joey00072/experiments-with-bitnet-1-5
NousResearch experiment NousResearch/OLMo-Bitnet-1B
🥳Tiny and large multimodal models great for embeddings
GitHub: https://github.com/unum-cloud/uform
Encoders: https://hf.co/collections/unum-cloud/multimodal-encoders-660553903617c5297eb16838
ONNX weights: https://hf.co/collections/unum-cloud/uform-vl-english-large-onnx-66055a57c182d846f3bc1949
📜 SMPLer-X: Expressive Human Pose and Shape Estimation
Project website: https://caizhongang.com/projects/SMPLer-X/
Demo: caizhongang/SMPLer-X
Paper: SMPLer-X: Scaling Up Expressive Human Pose and Shape Estimation (2309.17448)
🧙GeoWizard: 3D Geometry Estimation
Project website: https://fuxiao0719.github.io/projects/geowizard/
Demo: lemonaddie/geowizard
Misc models and datasets
- Dolphin-2.8-mistral-7b-v0.2 cognitivecomputations/dolphin-2.8-mistral-7b-v02
- Hermes-2-Pro-11B, a self-frankenmerge 11B variant mattshumer/Hermes-2-Pro-11B
- Large conversational dataset based on Usenet data in the Italian language mii-community/UsenetArchiveIT-conversations
-
- 3 replies
osanseviero
posted
an
update
12 months ago
Post
3558
Diaries of Open Source. Part 12 🤗
🚀Alibaba releases Qwen1.5-MoE-A2.7B, an interesting MoE with 2.7B activated parameters and 64 experts
Blog https://qwenlm.github.io/blog/qwen-moe/
Demo: Qwen/qwen1.5-MoE-A2.7B-Chat-demo
Models: https://hf.co/Qwen
GitHub: https://github.com/QwenLM/Qwen1.5
🎵VoiceCraft, SOTA speech editing and text to speech
GitHub: https://github.com/jasonppy/VoiceCraft
Model: pyp1/VoiceCraft
🐍 AI21Labs release Jamba, an SSM-Transformer, pretrained MoE which allows a large context window (256K) and high throughput
Blog https://www.ai21.com/blog/announcing-jamba
Model ai21labs/Jamba-v0.1
✨ Berkeley releases Starling-LM-7B, an RLHF-ed model, and -RM-34B, a Yi-based reward model very good for its size
Starling Beta: Nexusflow/Starling-LM-7B-beta
Starling RM: Nexusflow/Starling-RM-34B
🖥️Stability releases Stable Code Instruct 3B, an instruct model for code generation
Blog: https://stability.ai/news/introducing-stable-code-instruct-3b
Demo: stabilityai/stable-code-instruct-3b
Report: https://stability.ai/s/Stable_Code_TechReport_release.pdf
📚Common Corpus: the largest public domain dataset for training LLMs
Blog: https://hf.co/blog/Pclanglais/common-corpus
Dataset: https://hf.co/collections/PleIAs/common-corpus-65d46e3ea3980fdcd66a5613
Misc:
⚡GaLore: a very memory-efficient technique that allows pretraining models in consumer GPUs https://hf.co/blog/galore
Moirai
📈Moirai, foundation models for time series forecasting https://hf.co/collections/Salesforce/moirai-10-r-models-65c8d3a94c51428c300e0742
🔥 Mistral-ORPO-Capybara-7K, a high-quality Mistral fine-tune using ORPO, a new alignment technique kaist-ai/mistral-orpo-capybara-7k
🤯APISR, an anime super-resolution upscaling model HikariDawn/APISR
🚀Alibaba releases Qwen1.5-MoE-A2.7B, an interesting MoE with 2.7B activated parameters and 64 experts
Blog https://qwenlm.github.io/blog/qwen-moe/
Demo: Qwen/qwen1.5-MoE-A2.7B-Chat-demo
Models: https://hf.co/Qwen
GitHub: https://github.com/QwenLM/Qwen1.5
🎵VoiceCraft, SOTA speech editing and text to speech
GitHub: https://github.com/jasonppy/VoiceCraft
Model: pyp1/VoiceCraft
🐍 AI21Labs release Jamba, an SSM-Transformer, pretrained MoE which allows a large context window (256K) and high throughput
Blog https://www.ai21.com/blog/announcing-jamba
Model ai21labs/Jamba-v0.1
✨ Berkeley releases Starling-LM-7B, an RLHF-ed model, and -RM-34B, a Yi-based reward model very good for its size
Starling Beta: Nexusflow/Starling-LM-7B-beta
Starling RM: Nexusflow/Starling-RM-34B
🖥️Stability releases Stable Code Instruct 3B, an instruct model for code generation
Blog: https://stability.ai/news/introducing-stable-code-instruct-3b
Demo: stabilityai/stable-code-instruct-3b
Report: https://stability.ai/s/Stable_Code_TechReport_release.pdf
📚Common Corpus: the largest public domain dataset for training LLMs
Blog: https://hf.co/blog/Pclanglais/common-corpus
Dataset: https://hf.co/collections/PleIAs/common-corpus-65d46e3ea3980fdcd66a5613
Misc:
⚡GaLore: a very memory-efficient technique that allows pretraining models in consumer GPUs https://hf.co/blog/galore
Moirai
📈Moirai, foundation models for time series forecasting https://hf.co/collections/Salesforce/moirai-10-r-models-65c8d3a94c51428c300e0742
🔥 Mistral-ORPO-Capybara-7K, a high-quality Mistral fine-tune using ORPO, a new alignment technique kaist-ai/mistral-orpo-capybara-7k
🤯APISR, an anime super-resolution upscaling model HikariDawn/APISR
-
- 4 replies
osanseviero
posted
an
update
12 months ago
Post
2083
Diaries of Open Source. Part 11 🚀
🚀Databricks release DBRX, potentially the best open access model! A 132B Mixture of Experts with 36B active params and trained on 12 trillion tokens
Blog: https://www.databricks.com/blog/introducing-dbrx-new-state-art-open-llm
Base and instruct models: databricks/dbrx-6601c0852a0cdd3c59f71962
Demo: https://hf.co/spaces/databricks/dbrx-instruct
🤏1-bit and 2-bit quantization exploration using HQQ+
Blog post: https://mobiusml.github.io/1bit_blog/
Models: https://hf.co/collections/mobiuslabsgmbh/llama2-7b-hqq-6604257a96fc8b9c4e13e0fe
GitHub: https://github.com/mobiusml/hqq
📚Cosmopedia: a large-scale synthetic dataset for pre-training - it includes 25 billion tokens and 30 million files
Dataset: HuggingFaceTB/cosmopedia
Blog: https://hf.co/blog/cosmopedia
⭐Mini-Gemini: multi-modal VLMs, from 2B to 34B
Models: https://hf.co/collections/YanweiLi/mini-gemini-6603c50b9b43d044171d0854
Paper: Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models (2403.18814)
GitHub: https://github.com/dvlab-research/MiniGemini
🔥VILA - On Pre-training for VLMs
Models: Efficient-Large-Model/vila-on-pre-training-for-visual-language-models-65d8022a3a52cd9bcd62698e
Paper: VILA: On Pre-training for Visual Language Models (2312.07533)
Misc
👀 FeatUp: a framework for image features at any resolution: mhamilton723/FeatUp FeatUp: A Model-Agnostic Framework for Features at Any Resolution (2403.10516)
🍞ColBERTus Maxiums, a colbertialized embedding model mixedbread-ai/mxbai-colbert-large-v1
🖌️Semantic Palette, a new drawing paradigm ironjr/SemanticPalette
🧑⚕️HistoGPT, a vision model that generates accurate pathology reports marr-peng-lab/histogpt https://www.medrxiv.org/content/10.1101/2024.03.15.24304211v1
🚀Databricks release DBRX, potentially the best open access model! A 132B Mixture of Experts with 36B active params and trained on 12 trillion tokens
Blog: https://www.databricks.com/blog/introducing-dbrx-new-state-art-open-llm
Base and instruct models: databricks/dbrx-6601c0852a0cdd3c59f71962
Demo: https://hf.co/spaces/databricks/dbrx-instruct
🤏1-bit and 2-bit quantization exploration using HQQ+
Blog post: https://mobiusml.github.io/1bit_blog/
Models: https://hf.co/collections/mobiuslabsgmbh/llama2-7b-hqq-6604257a96fc8b9c4e13e0fe
GitHub: https://github.com/mobiusml/hqq
📚Cosmopedia: a large-scale synthetic dataset for pre-training - it includes 25 billion tokens and 30 million files
Dataset: HuggingFaceTB/cosmopedia
Blog: https://hf.co/blog/cosmopedia
⭐Mini-Gemini: multi-modal VLMs, from 2B to 34B
Models: https://hf.co/collections/YanweiLi/mini-gemini-6603c50b9b43d044171d0854
Paper: Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models (2403.18814)
GitHub: https://github.com/dvlab-research/MiniGemini
🔥VILA - On Pre-training for VLMs
Models: Efficient-Large-Model/vila-on-pre-training-for-visual-language-models-65d8022a3a52cd9bcd62698e
Paper: VILA: On Pre-training for Visual Language Models (2312.07533)
Misc
👀 FeatUp: a framework for image features at any resolution: mhamilton723/FeatUp FeatUp: A Model-Agnostic Framework for Features at Any Resolution (2403.10516)
🍞ColBERTus Maxiums, a colbertialized embedding model mixedbread-ai/mxbai-colbert-large-v1
🖌️Semantic Palette, a new drawing paradigm ironjr/SemanticPalette
🧑⚕️HistoGPT, a vision model that generates accurate pathology reports marr-peng-lab/histogpt https://www.medrxiv.org/content/10.1101/2024.03.15.24304211v1
-
-
- 4 replies
osanseviero
posted
an
update
12 months ago
Post
1625
Diaries of Open Source. Part 10 🚀
🌼Marigold-LCM: A super fast SOTA Depth Estimator
Demo: prs-eth/marigold-lcm
Original paper: Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation (2312.02145)
Model: https://hf.co/prs-eth/marigold-lcm-v1-0
🌟Quiet-STaR: A self-teaching technique via internal monologue
Paper: Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking (2403.09629)
GitHub: https://github.com/ezelikman/quiet-star
Tweetutorial: https://twitter.com/ericzelikman/status/1768663835106513041
🖼️ WebSight v0.2: A image-to-code dataset containing tailwind CSS, images in screenshots, and more!
Dataset: HuggingFaceM4/WebSight
Paper: Unlocking the conversion of Web Screenshots into HTML Code with the WebSight Dataset (2403.09029)
Blog: https://hf.co/blog/websight
🕵️Agent-FLAN - effective agent tuning for LLMs
Paper: Agent-FLAN: Designing Data and Methods of Effective Agent Tuning for Large Language Models (2403.12881)
Model: internlm/Agent-FLAN-7b
Dataset: internlm/Agent-FLAN
Website: https://internlm.github.io/Agent-FLAN/
🔥HPT, a family of multimodal LLMs from HyperGAI
Blog post: https://hypergai.com/blog/introducing-hpt-a-family-of-leading-multimodal-llms
Model: HyperGAI/HPT
GitHub: https://github.com/hyperGAI/HPT
🌏Models and datasets around the world
- Tess-70B, a MiQu-70B fine-tune with high-quality data migtissera/Tess-70B-v1.6
- UNI, a model trained on 100 million pathology images from 100k+ slides MahmoodLab/UNI
- CONCH, a VLM trained on 1.17 million pathology image-text pairs MahmoodLab/CONCH
🌼Marigold-LCM: A super fast SOTA Depth Estimator
Demo: prs-eth/marigold-lcm
Original paper: Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation (2312.02145)
Model: https://hf.co/prs-eth/marigold-lcm-v1-0
🌟Quiet-STaR: A self-teaching technique via internal monologue
Paper: Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking (2403.09629)
GitHub: https://github.com/ezelikman/quiet-star
Tweetutorial: https://twitter.com/ericzelikman/status/1768663835106513041
🖼️ WebSight v0.2: A image-to-code dataset containing tailwind CSS, images in screenshots, and more!
Dataset: HuggingFaceM4/WebSight
Paper: Unlocking the conversion of Web Screenshots into HTML Code with the WebSight Dataset (2403.09029)
Blog: https://hf.co/blog/websight
🕵️Agent-FLAN - effective agent tuning for LLMs
Paper: Agent-FLAN: Designing Data and Methods of Effective Agent Tuning for Large Language Models (2403.12881)
Model: internlm/Agent-FLAN-7b
Dataset: internlm/Agent-FLAN
Website: https://internlm.github.io/Agent-FLAN/
🔥HPT, a family of multimodal LLMs from HyperGAI
Blog post: https://hypergai.com/blog/introducing-hpt-a-family-of-leading-multimodal-llms
Model: HyperGAI/HPT
GitHub: https://github.com/hyperGAI/HPT
🌏Models and datasets around the world
- Tess-70B, a MiQu-70B fine-tune with high-quality data migtissera/Tess-70B-v1.6
- UNI, a model trained on 100 million pathology images from 100k+ slides MahmoodLab/UNI
- CONCH, a VLM trained on 1.17 million pathology image-text pairs MahmoodLab/CONCH
-
- 4 replies
osanseviero
posted
an
update
12 months ago
Post
3276
Diaries of Open Source. Part 9!
⏰Amazon releases Chronos, a family of models for time series
Base model: amazon/chronos-t5-large
Paper: Chronos: Learning the Language of Time Series (2403.07815)
Models: https://huggingface.co./collections/amazon/chronos-models-65f1791d630a8d57cb718444
💡ORPO Alignment: align without a reference model nor SFT!
Paper: ORPO: Monolithic Preference Optimization without Reference Model (2403.07691)
Models: kaist-ai/orpo-65efef87544ba100aef30013
GitHub: https://github.com/xfactlab/orpo
🇺🇳Cohere releases 250M Wikipedia Embeddings in 300+ languages
Data: Cohere/wikipedia-2023-11-embed-multilingual-v3
Announcement: https://twitter.com/Nils_Reimers/status/1767891859207057618
🧬SegmentNT: a LLM for annotating DNA at single nucleotide resolution
Models: InstaDeepAI/segmentnt-65eb4941c57808b4a3fe1319
GitHub repo: https://github.com/instadeepai/nucleotide-transformer
Paper: https://www.biorxiv.org/content/10.1101/2024.03.14.584712v1
🚀DynamiCrafter: video generation models for interpolation and looping are out!
Project page: https://doubiiu.github.io/projects/DynamiCrafter/
GitHub: https://github.com/Doubiiu/DynamiCrafter
Demo: Doubiiu/DynamiCrafter_interp_loop
🚀Stanford releases Anticipatory Music Transformer:
GitHub: https://github.com/jthickstun/anticipation/
Models: https://hf.co/stanford-crfm
Original blog announcement: https://crfm.stanford.edu/2023/06/16/anticipatory-music-transformer.html
⏰Amazon releases Chronos, a family of models for time series
Base model: amazon/chronos-t5-large
Paper: Chronos: Learning the Language of Time Series (2403.07815)
Models: https://huggingface.co./collections/amazon/chronos-models-65f1791d630a8d57cb718444
💡ORPO Alignment: align without a reference model nor SFT!
Paper: ORPO: Monolithic Preference Optimization without Reference Model (2403.07691)
Models: kaist-ai/orpo-65efef87544ba100aef30013
GitHub: https://github.com/xfactlab/orpo
🇺🇳Cohere releases 250M Wikipedia Embeddings in 300+ languages
Data: Cohere/wikipedia-2023-11-embed-multilingual-v3
Announcement: https://twitter.com/Nils_Reimers/status/1767891859207057618
🧬SegmentNT: a LLM for annotating DNA at single nucleotide resolution
Models: InstaDeepAI/segmentnt-65eb4941c57808b4a3fe1319
GitHub repo: https://github.com/instadeepai/nucleotide-transformer
Paper: https://www.biorxiv.org/content/10.1101/2024.03.14.584712v1
🚀DynamiCrafter: video generation models for interpolation and looping are out!
Project page: https://doubiiu.github.io/projects/DynamiCrafter/
GitHub: https://github.com/Doubiiu/DynamiCrafter
Demo: Doubiiu/DynamiCrafter_interp_loop
🚀Stanford releases Anticipatory Music Transformer:
GitHub: https://github.com/jthickstun/anticipation/
Models: https://hf.co/stanford-crfm
Original blog announcement: https://crfm.stanford.edu/2023/06/16/anticipatory-music-transformer.html
-
- 2 replies
osanseviero
posted
an
update
12 months ago
Post
2536
Diaries of Open Source. Part 8!
🤯CRM: Image-to-3D Textured Mesh
Demo: Zhengyi/CRM
Model: Zhengyi/CRM
Project page: https://ml.cs.tsinghua.edu.cn/~zhengyi/CRM/
Paper: CRM: Single Image to 3D Textured Mesh with Convolutional Reconstruction Model (2403.05034)
🤏Half Quadratic Quantization: super-fast quantization of very large models
Blog post: https://mobiusml.github.io/hqq_blog/
Colab: https://colab.research.google.com/drive/1cG_5R_u9q53Uond7F0JEdliwvoeeaXVN?usp=sharing
Repo: https://github.com/mobiusml/hqq
🤗GemMoE -Gemma + MoE
Model: Crystalcareai/GemMoE-Base-Random
Collection: Crystalcareai/gemmoe-65f11f4922af97ebe9943591
👀VeCLIP and MOFI, new 0-shot and image retrieval models by Apple, are now open-source!
GitHub: https://github.com/apple/ml-veclip/ and https://github.com/apple/ml-mofi
VeCLIP paper: From Scarcity to Efficiency: Improving CLIP Training via Visual-enriched Captions (2310.07699)
MOFI paper: MOFI: Learning Image Representations from Noisy Entity Annotated Images (2306.07952)
⚡SPIN: Recipe for alignment with very little data
Collection: argilla/dibt-prompt-collective-spin-65ef59062518776024395fc3
Tweetutorial: https://twitter.com/argilla_io/status/1767608154697699455
👀ViT Prisma - an interoperability library for vision models
GitHub: https://github.com/soniajoseph/ViT-Prisma
☕OpenLRM: full model and training code are open-sourced
Codebase: https://github.com/3DTopia/OpenLRM
Demo: zxhezexin/OpenLRM
Models: https://huggingface.co./zxhezexin
⚗️Oxford releases an extensive PEFT evaluation for bio models
Model: NTaylor/bio-mobilebert-mimic-mp-lora
GitHub: https://github.com/nlpie-research/efficient-ml
Paper: Efficiency at Scale: Investigating the Performance of Diminutive Language Models in Clinical Tasks (2402.10597)
🌍Data and models around the world
Hermes 2 Pro 7B: an upgraded Nous Hermes 2 model with strong function calling and JSON capabilities NousResearch/Hermes-2-Pro-Mistral-7B
Navarasa-2.0 : Gemma fine-tuned in 15 indian language Telugu-LLM-Labs/navarasa-65f5e6ffdf29f02c6d7767ce
🤯CRM: Image-to-3D Textured Mesh
Demo: Zhengyi/CRM
Model: Zhengyi/CRM
Project page: https://ml.cs.tsinghua.edu.cn/~zhengyi/CRM/
Paper: CRM: Single Image to 3D Textured Mesh with Convolutional Reconstruction Model (2403.05034)
🤏Half Quadratic Quantization: super-fast quantization of very large models
Blog post: https://mobiusml.github.io/hqq_blog/
Colab: https://colab.research.google.com/drive/1cG_5R_u9q53Uond7F0JEdliwvoeeaXVN?usp=sharing
Repo: https://github.com/mobiusml/hqq
🤗GemMoE -Gemma + MoE
Model: Crystalcareai/GemMoE-Base-Random
Collection: Crystalcareai/gemmoe-65f11f4922af97ebe9943591
👀VeCLIP and MOFI, new 0-shot and image retrieval models by Apple, are now open-source!
GitHub: https://github.com/apple/ml-veclip/ and https://github.com/apple/ml-mofi
VeCLIP paper: From Scarcity to Efficiency: Improving CLIP Training via Visual-enriched Captions (2310.07699)
MOFI paper: MOFI: Learning Image Representations from Noisy Entity Annotated Images (2306.07952)
⚡SPIN: Recipe for alignment with very little data
Collection: argilla/dibt-prompt-collective-spin-65ef59062518776024395fc3
Tweetutorial: https://twitter.com/argilla_io/status/1767608154697699455
👀ViT Prisma - an interoperability library for vision models
GitHub: https://github.com/soniajoseph/ViT-Prisma
☕OpenLRM: full model and training code are open-sourced
Codebase: https://github.com/3DTopia/OpenLRM
Demo: zxhezexin/OpenLRM
Models: https://huggingface.co./zxhezexin
⚗️Oxford releases an extensive PEFT evaluation for bio models
Model: NTaylor/bio-mobilebert-mimic-mp-lora
GitHub: https://github.com/nlpie-research/efficient-ml
Paper: Efficiency at Scale: Investigating the Performance of Diminutive Language Models in Clinical Tasks (2402.10597)
🌍Data and models around the world
Hermes 2 Pro 7B: an upgraded Nous Hermes 2 model with strong function calling and JSON capabilities NousResearch/Hermes-2-Pro-Mistral-7B
Navarasa-2.0 : Gemma fine-tuned in 15 indian language Telugu-LLM-Labs/navarasa-65f5e6ffdf29f02c6d7767ce
-
-
- 4 replies