open-llm-leaderboard/Nexesenex__Llama_3.1_8b_Stormeder_v1.04-details
Updated
Open LLM Leaderboard
Enterprise
community
AI & ML interests
Evaluating open LLMs
Recent Activity
View all activity
open-llm-leaderboard's activity
open-llm-botΒ
updated
2
datasets
less than a minute ago
open-llm-botΒ
published
a
dataset
1 minute ago
open-llm-botΒ
updated
a
dataset
1 minute ago
open-llm-botΒ
published
a
dataset
2 minutes ago
open-llm-botΒ
updated
a
dataset
30 minutes ago
open-llm-botΒ
updated
a
dataset
31 minutes ago
Post
102
The AI race in the automotive industry is heating upπ
Li Autoβs research team has released their latest paper on LLMπ LDGen: Enhancing Text-to-Image Synthesis via Large Language Model-Driven Language Representation (2502.18302)
β¨This paper introduces LDGen, which integrates LLMs with diffusion models to enhance text-to-image (T2I) generation capabilities.
Li Autoβs research team has released their latest paper on LLMπ LDGen: Enhancing Text-to-Image Synthesis via Large Language Model-Driven Language Representation (2502.18302)
β¨This paper introduces LDGen, which integrates LLMs with diffusion models to enhance text-to-image (T2I) generation capabilities.
Post
94
LLaDA π₯a 8B diffusion model by GSAI Lab Renmin University
β¨Fully trained from scratch, LLaDA delivers performance on par with LLaMA3 8B
Model: GSAI-ML/LLaDA-8B-Instruct
Demo: multimodalart/LLaDA
Paper: Large Language Diffusion Models (2502.09992)
β¨Fully trained from scratch, LLaDA delivers performance on par with LLaMA3 8B
Model: GSAI-ML/LLaDA-8B-Instruct
Demo: multimodalart/LLaDA
Paper: Large Language Diffusion Models (2502.09992)
freddyaboultonΒ
posted
an
update
3 days ago
Post
2819
Getting WebRTC and Websockets right in python is very tricky. If you've tried to wrap an LLM in a real-time audio layer then you know what I'm talking about.
That's where FastRTC comes in! It makes WebRTC and Websocket streams super easy with minimal code and overhead.
Check out our org: hf.co/fastrtc
That's where FastRTC comes in! It makes WebRTC and Websocket streams super easy with minimal code and overhead.
Check out our org: hf.co/fastrtc
Post
2574
Wan2.1 π₯πΉ new OPEN video model by Alibaba Wan team!
Model: Wan-AI/Wan2.1-T2V-14B
Demo: Wan-AI/Wan2.1
β¨Apache 2.0
β¨8.19GB VRAM, runs on most GPUs
β¨Multi-Tasking: T2V, I2V, Video Editing, T2I, V2A
β¨Text Generation: Supports Chinese & English
β¨Powerful Video VAE: Encode/decode 1080P w/ temporal precision
Model: Wan-AI/Wan2.1-T2V-14B
Demo: Wan-AI/Wan2.1
β¨Apache 2.0
β¨8.19GB VRAM, runs on most GPUs
β¨Multi-Tasking: T2V, I2V, Video Editing, T2I, V2A
β¨Text Generation: Supports Chinese & English
β¨Powerful Video VAE: Encode/decode 1080P w/ temporal precision
- 1 reply
Post
2822
Try QwQ-Max-Preview, Qwen's reasoning model hereπ https://chat.qwen.ai
Can't wait for the model weights to drop on the Hugging Face Hub π₯
Can't wait for the model weights to drop on the Hugging Face Hub π₯
- 2 replies
Post
2395
Two AI startups, DeepSeek & Moonshot AI , keep moving in perfect sync π
β¨ Last December: DeepSeek & Moonshot AI released their reasoning models on the SAME DAY.
DeepSeek: deepseek-ai/DeepSeek-R1
MoonShot: https://github.com/MoonshotAI/Kimi-k1.5
β¨ Last week: Both teams published papers on modifying attention mechanisms on the SAME DAY AGAIN.
DeepSeek: Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention (2502.11089)
Moonshot: MoBA: Mixture of Block Attention for Long-Context LLMs (2502.13189)
β¨ TODAY:
DeepSeek unveiled Flash MLA: a efficient MLA decoding kernel for NVIDIA Hopper GPUs, optimized for variable-length sequences.
https://github.com/deepseek-ai/FlashMLA
Moonshot AI introduces Moonlight: a 3B/16B MoE trained on 5.7T tokens using Muon, pushing the Pareto frontier with fewer FLOPs.
moonshotai/Moonlight-16B-A3B
What's next? π
β¨ Last December: DeepSeek & Moonshot AI released their reasoning models on the SAME DAY.
DeepSeek: deepseek-ai/DeepSeek-R1
MoonShot: https://github.com/MoonshotAI/Kimi-k1.5
β¨ Last week: Both teams published papers on modifying attention mechanisms on the SAME DAY AGAIN.
DeepSeek: Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention (2502.11089)
Moonshot: MoBA: Mixture of Block Attention for Long-Context LLMs (2502.13189)
β¨ TODAY:
DeepSeek unveiled Flash MLA: a efficient MLA decoding kernel for NVIDIA Hopper GPUs, optimized for variable-length sequences.
https://github.com/deepseek-ai/FlashMLA
Moonshot AI introduces Moonlight: a 3B/16B MoE trained on 5.7T tokens using Muon, pushing the Pareto frontier with fewer FLOPs.
moonshotai/Moonlight-16B-A3B
What's next? π
Post
749
VLM-R1π₯bringing DeepSeekβs R1 method to vision language models!
GitHub: https://github.com/om-ai-lab/VLM-R1
Demo: omlab/VLM-R1-Referral-Expression
GitHub: https://github.com/om-ai-lab/VLM-R1
Demo: omlab/VLM-R1-Referral-Expression
Post
4179
π StepFunιΆθ·ζθΎ° is making BIG open moves!
Last year, their GOT-OCR 2.0 took the community by storm π₯but many didnβt know they were also building some amazing models. Now, theyβve just dropped something huge on the hub!
πΊ Step-Video-T2V: a 30B bilingual open video model that generates 204 frames (8-10s) at 540P resolution with high information density & consistency.
stepfun-ai/stepvideo-t2v
π Step-Audio-TTS-3B : a TTS trained with the LLM-Chat paradigm on a large synthetic dataset, capable of generating RAP & Humming
stepfun-ai/step-audio-67b33accf45735bb21131b0b
Last year, their GOT-OCR 2.0 took the community by storm π₯but many didnβt know they were also building some amazing models. Now, theyβve just dropped something huge on the hub!
πΊ Step-Video-T2V: a 30B bilingual open video model that generates 204 frames (8-10s) at 540P resolution with high information density & consistency.
stepfun-ai/stepvideo-t2v
π Step-Audio-TTS-3B : a TTS trained with the LLM-Chat paradigm on a large synthetic dataset, capable of generating RAP & Humming
stepfun-ai/step-audio-67b33accf45735bb21131b0b
-
- 3 replies
Post
2420
The latest paper of DeepSeek is now available on the Daily Papers page π
You can reach out to the authors directly on this pageπ
Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention (2502.11089)
You can reach out to the authors directly on this pageπ
Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention (2502.11089)
- 1 reply
Post
2553
Ovis2 π₯ a multimodal LLM released by Alibaba AIDC team.
AIDC-AI/ovis2-67ab36c7e497429034874464
β¨1B/2B/4B/8B/16B/34B
β¨Strong CoT for deeper problem solving
β¨Multilingual OCR β Expanded beyond English & Chinese, with better data extraction
AIDC-AI/ovis2-67ab36c7e497429034874464
β¨1B/2B/4B/8B/16B/34B
β¨Strong CoT for deeper problem solving
β¨Multilingual OCR β Expanded beyond English & Chinese, with better data extraction
Post
3547
InspireMusic π΅π₯ an open music generation framework by Alibaba FunAudio Lab
Model: FunAudioLLM/InspireMusic-1.5B-Long
Demo: FunAudioLLM/InspireMusic
β¨ Music, songs, audio - ALL IN ONE
β¨ High quality audio: 24kHz & 48kHz sampling rates
β¨ Long-Form Generation: enables extended audio creation
β¨ Efficient Fine-Tuning: precision (BF16, FP16, FP32) with user-friendly scripts
Model: FunAudioLLM/InspireMusic-1.5B-Long
Demo: FunAudioLLM/InspireMusic
β¨ Music, songs, audio - ALL IN ONE
β¨ High quality audio: 24kHz & 48kHz sampling rates
β¨ Long-Form Generation: enables extended audio creation
β¨ Efficient Fine-Tuning: precision (BF16, FP16, FP32) with user-friendly scripts
-
- 1 reply
Post
4598
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 πͺ
Whatβs new compared to existing reasoning datasets?
βΎ 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.
π³ 800k R1 reasoning traces: We generate two answers for 400k problems using DeepSeek R1. The filtered dataset contains 220k problems with correct reasoning traces.
π 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.
β³ 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 canβt be verified with a rules-based parser)
π We match the performance of DeepSeek-Distill-Qwen-7B by finetuning Qwen-7B-Math-Instruct on our dataset.
π 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 πͺ
Whatβs new compared to existing reasoning datasets?
βΎ 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.
π³ 800k R1 reasoning traces: We generate two answers for 400k problems using DeepSeek R1. The filtered dataset contains 220k problems with correct reasoning traces.
π 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.
β³ 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 canβt be verified with a rules-based parser)
π We match the performance of DeepSeek-Distill-Qwen-7B by finetuning Qwen-7B-Math-Instruct on our dataset.
π Read our blog post for all the nitty gritty details: https://huggingface.co./blog/open-r1/update-2
hynkyΒ
authored
a
paper
22 days ago