morrowind

joined 3 weeks ago
sorted by: new top controversial old
EXAONE Deep ━ Setting a New Standard for Reasoning AI - LG AI Research News in c/localllama@sh.itjust.works
[–] morrowind@lemm.ee 1 points 4 days ago (1 children)

what is the license? The link on hf just 404s

4
EXAONE Deep ━ Setting a New Standard for Reasoning AI - LG AI Research News (www.lgresearch.ai)
Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching in c/localllama@sh.itjust.works
[–] morrowind@lemm.ee 2 points 5 days ago

Very similar to chain of draft but seems more thorough

12
Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching (arxiv.org)
5
Sorting-Free GPU Kernels for LLM Sampling (flashinfer.ai)
Reka Flash, open source 21B model comparable to QWQ 32B in c/localllama@sh.itjust.works
18
Reka Flash, open source 21B model comparable to QWQ 32B (i.postimg.cc)
 
Qwen/QwQ-32B · Hugging Face in c/localllama@sh.itjust.works
[–] morrowind@lemm.ee 3 points 2 weeks ago

It matches R1 in the given benchmarks. R1 has 671B params (36 activated) while this only has 32

Qwen/QwQ-32B · Hugging Face in c/localllama@sh.itjust.works
[–] morrowind@lemm.ee 2 points 2 weeks ago (2 children)

insane, absolutely insane

8
Chain of Draft: Thinking Faster by Writing Less (arxiv.org)
12
Atom of Thoughts (AOT): lifts gpt-4o-mini to 80.6% F1 on HotpotQA, surpassing o3-mini and DeepSeek-R1 (bsky.app)
Alibaba Releases Advanced Open Video Model, Immediately Becomes AI Porn Machine in c/technology@lemmy.world
[–] morrowind@lemm.ee 6 points 3 weeks ago (2 children)

good luck trying to run a video model locally

Unless you have top tier hardware