←
Yurko117 posts
Yurko avatar

Yurko

@DJTechDebt

Here mostly to read news.

◷ Joined March 2014▧ 8 Photos and videos
José Valim avatar
⟳ Yurko retweeted
José Valim@josevalim·Jun 3

Elixir v1.20 released! Now officially a gradually typed language: Elixir type checks every single line of code, finding bugs and dead code, without developer overhead (no typing signatures) and extremely low false positives rate. Plus a faster compiler! Links and reports below.

Yurko avatar
Yurko@DJTechDebt·Mar 4

New Qwen 3.5 lineup is nothing short of breakthrough! Genuinely impressed with coherence, knowledge and skill that are available on consumer hardware courtesy of @JustinLin610 and the team. All the best in your next adventure! Below are some fun 27b replies.

Post attachmentPost attachment
Yurko avatar
Yurko@DJTechDebt·19 Oct 2025

Qwen3 4b VL is unexpectedly good for its size (trying it out in LM Studio) - so far it was able to count people in the picture (15), OCR verbatim, read and transcribe the table of data in Ukrainian - impressive!

vik avatar
⟳ Yurko retweeted
vik@vikhyatk·18 Sep 2025

Excited to release a preview of Moondream 3. A 9B param, 2B active MoE vision language model that makes no compromises; offering state-of-the-art visual reasoning while still retaining an efficient and deployment-friendly form factor.

0:48
Junyang Lin avatar
⟳ Yurko retweeted
Junyang Lin@JustinLin610·30 Jul 2025

these two days we are sharing the instruct and thinking models of our smaller variant of the 2507 seires, 30a3-2507. fast, but much smarter than before. i like this size, it is just something that i can easily play with, which is also somehow smart enough. btw, i hope we can shoot coder 30a3 tmr.

Qwen@Alibaba_Qwen30 Jul 2025

🚀 Qwen3-30B-A3B-Thinking-2507, a medium-size model that can think! • Nice performance on reasoning tasks, including math, science, code & beyond • Good at tool use, competitive with larger models • Native support of 256K-token context, extendable to 1M Qwen Chat: Go to  chat.qwen.ai/?model=Qwen3-30…  and click “Thinking”. HF:huggingface.co/Qwen/Qwen3-30… or huggingface.co/Qwen/Qwen3-30… ModelScope: modelscope.cn/models/Qwen/Qw… or modelscope.cn/models/Qwen/Qw…

Post attachment
Zach Daniel avatar
⟳ Yurko retweeted
Zach Daniel@ZachSDaniel1·24 Jun 2025

We need a way to validate approaches to how we work with LLMs to write code, and we need to start developing the dataset. I've laid some groundwork here, but I need help. Tear it up, critique it, but most of all, PR some evals y'all. #MyElixirStatus github.com/ash-project/evals

Zach Daniel avatar
⟳ Yurko retweeted
Zach Daniel@ZachSDaniel1·9 Jun 2025

Elixir 1.19 is a banger! I'm so pleased with the direction that #ElixirLang is going. My programs just get faster and more correct every time. I just know that we're in good hands. Thank you to everyone on the team for your hard work! github.com/elixir-lang/elixi… #MyElixirStatus

Nous Research avatar
⟳ Yurko retweeted
Nous Research@NousResearch·14 May 2025

We are launching testnet with the pre-training of a 40B parameter LLM: - MLA Architecture - Dataset consisting of FineWeb (14T) + FineWeb-2 minus some less common languages (4T), and The Stack v2 (1T) The resulting model will be small enough to train on with a single H/DGX and run on a 3090, but it will be powerful enough to serve as the basis for strong reasoning models and creative pursuits. The model will be trained continually without a final "annealing" step, resulting in a true unaltered base model. Looking ahead, we will draw model ideas from the community via our forum and Discord. By enabling highly parallel and scalable experimentation, we’re betting that the next innovation in model creation and design will come from the open source community.

Post attachment
Paulo Valente avatar
⟳ Yurko retweeted
Paulo Valente@polvalente·11 May 2025

I've just released Handoff, a new Elixir library for distributed function graph execution. hexdocs.pm/handoff Among other things, the library enables distributed Nx computations (anyone said distributed inference?) A huge part of this was sponsored by @tvlabs!