OPEN SOURCE - AI Dynamics

Hermes 3 Llama 3.2 3B Fine-Tuned for Coding

By

–

11 December 2024 22h29

Nous dropped a Hermes 3 Llama 3.2 3B Full fine-tune of L3.2 3B focused on coding, structured outputs and function calling! Handles long-context, multi-turn conversations and reasoning brilliantly too

→ View original post on X — @reach_vb

11 December 2024

Sustainable Funding Models for Open Source Technology Projects

By

@mjasay

–

11 December 2024 20h35

So charge for it, whether directly to users or through corporate sponsorships. I've seen this movie before in open source: relying on the charity of others to keep tech running doesn't work. Even Mozilla figured out a way to pay for its existence (ie antitrust shield for Google)

→ View original post on X — @mjasay

11 December 2024

DuckDB 1.0: High-Speed Analytics Engine for AI Workflows

By

@mikequindazzi

–

11 December 2024 19h17

#DuckDB Redefining Analytics for #AI 100x faster Parquet queries vs. Pandas Process billions of rows in seconds In-process #SQL <10MB RAM for setup Integrated #Python/#R for #ML workflows Version 1.0 stability + @MotherDuckHQ backing #DataAnalytics #TechTrends

→ View original post on X — @mikequindazzi

11 December 2024

Llama 3.3 70B Now Available on SambaNova Cloud

By

@sambanovaai

–

11 December 2024 18h43

🚨 Llama 3.3 70B, now available on SambaNova Cloud! @AIatMeta

✅ Context length of 64K tokens
✅ Perfect for building Agents with other models
✅ Need higher rate limits? Just ask us

Try it out 👇
— SambaNova (@SambaNovaAI) 11 décembre 2024

Llama 3.3 70B, now available on SambaNova Cloud! @AIatMeta Context length of 64K tokens Perfect for building Agents with other models Need higher rate limits? Just ask us Try it out

→ View original post on X — @sambanovaai

11 December 2024

Meta Llama 3.3 integration and bug fixes release

By

@flowiseai

–

11 December 2024 17h53

We're also bringing in Llama 3.3 by Meta, and many bugfixes Full release note:

→ View original post on X — @flowiseai

11 December 2024

Flash 8B Release: Community Request for Previous Model Access

By

@reach_vb

–

11 December 2024 16h55

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!

→ View original post on X — @reach_vb

11 December 2024

Torch.compile() Outperforms Triton for Kernel Optimization

By

@alexjc

–

11 December 2024 12h55

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.

→ View original post on X — @alexjc

11 December 2024