Congrats on the release! Are there plans to release the previous Flash 8B? I think it'll be a really helpful research artefact for the community!
OPEN SOURCE
-
Torch.compile() Outperforms Triton for Kernel Optimization
By
–
It's interesting that this "kernel" is designed for torch.compile(), so it's Python code but turns out faster! Conversely, some of the other Liger kernels are Triton and I measured them as slower than torch.compile'd versions.
-
Open LLM Leaderboard Gets New Frontend Interface
By
–
The Open LLM Leaderboard got a new front for Christmas
-
Top Trending Text Generation Models on Hugging Face
By
–
This should be the most up-to-date: https://
huggingface.co/models?inferen
ce=warm&pipeline_tag=text-generation&sort=trending
… -
Devin Agent Contributing Code to MCP Repository
By
–
Devin (powered by Anthropic models) contributing PRs to the MCP repo Agents improving agents, it only moves faster from here on out
-
DeepSeek-V2.5-1210 with Live Search Support
By
–
BREAKING : DeepSeek got a model upgrade to DeepSeek-V2.5-1210 along with live search support A new Search toggle is now available on the prompt bar.
-
Flux Open-Source Model: Generation Quality, Parameters & Prompt Optimization
By
–
Flux is an open-source model that excels in creating realistic human portraits, product photography, UI design templates, and more.
— Satya Mallick (@LearnOpenCV) 10 décembre 2024
In this article, we'll explore the model’s generation quality, limitations, and the impact of various prompt styles by tuning its key parameters.… pic.twitter.com/7SRNPAAlarFlux is an open-source model that excels in creating realistic human portraits, product photography, UI design templates, and more. In this article, we'll explore the model’s generation quality, limitations, and the impact of various prompt styles by tuning its key parameters.
-
DeepSeek-V2.5-1210 Model Weights Released on Hugging Face
By
–
Model weights: https://
huggingface.co/deepseek-ai/De
epSeek-V2.5-1210
… -
DeepSeek-V2.5-1210 Upgrade Achieves 82.8% on MATH-500
By
–
Let’s gooo! The whale is back w/ DeepSeek-V2.5-1210 an upgraded version of DeepSeek-V2.5, offering improvements in: > 74.8% to 82.8% on MATH-500
> 29.2% to 34.38% on LiveCodebench
> writing and Reasoning: notable improvements in internal tests
> optimised file upload and -
Alibaba Qwen 2.5-32B-Coder: Fast Open-Source Model Release
By
–
📣 @Alibaba_Qwen 2.5-32B-Coder now available!
— SambaNova (@SambaNovaAI) 9 décembre 2024
✅ Access the best open-source models
✅ 5X faster than any #GPU provider
We've partnered with @continuedev to make it easier for #devs to integrate these models & experience fast #inference response times.
See how: 👇@Alibaba_Qwen 2.5-32B-Coder now available! Access the best open-source models 5X faster than any #GPU provider We've partnered with @continuedev to make it easier for #devs to integrate these models & experience fast #inference response times. See how: