codelion

10 ms·

121.
Unbounded Context with Memory
1 points by codelion 2y ago|1 comments
122.
by codelion 2y ago
This is surprising to only those that have not worked in formal reasoning. Yes, LLMs cannot do true logical reasoning in a formal sense, you can do better with an SMT solver. But it is also true that you can solve a lot of logical problems
123.
Optillm: An Optimizing Inference Proxy with Plugins(github.com)
2 points by codelion 2y ago|1 comments
124.
by codelion 2y ago
Optillm is an optimizing inference proxy that has over a dozen techniques that aim to improve the accuracy of the responses using test-time compute. Over the last couple of months we have set several SOTA results using smaller and less capa
125.
Discover LLM-assisted workflow opportunities in your code(patched.codes)
1 points by codelion 2y ago|0 comments
126.
Build your own AutoFix with Patchwork(patched.codes)
12 points by codelion 2y ago|0 comments
127.
Cot Decoding – Eliciting Reasoning from LLMs
2 points by codelion 2y ago|0 comments
128.
Evaluating code-to-readme generation using LLMs(patched.codes)
3 points by codelion 2y ago|0 comments
129.
by codelion 2y ago
That’s the hardest part, figuring out the reward. For generic tasks it is not easy, in my implementation in optillm I am using the llm itself to generate a score based on the mcts trajectory. But that is not as good as having a reward that
130.
by codelion 2y ago
This is very cool, I had worked on something similar back at https://securade.ai using a Nvidia Jetson as the edge device. I am now excited about reCamera - https://www.seeedstudio.com/blog/recamera/
131.
by codelion 2y ago
This is good I also had worked on something similar in optillm - https://github.com/codelion/optillm . You can do this with any LLM and several optimization techniques (including cot_reflection) like mcts, plansearch, m
132.
by codelion 2y ago
I have also spent some time on 2) and implemented several approaches in this open source optimising llm proxy - https://github.com/codelion/optillm In my experience it does work quite well, but we probably need differe
133.
by codelion 2y ago
You have just listed the benefits of having a CLA. This post is about why projects should pick a license not how contributors should prioritise what they spend their time on. If you are trying to build something commercial and be open-sourc
134.
by codelion 2y ago
I started this thread on the w3c list almost 20 years ago - https://lists.w3.org/Archives/Public/semantic-web/2005Dec/00... Unfortunately, it is unlikely we will ever get something like a Semantic web. I
135.
by codelion 2y ago
What's wrong in having a CLA? It allows you to re-license your project in future.
136.
by codelion 2y ago
Is there a hosted version one can try without having to set it up locally?
137.
by codelion 2y ago
This resonated with us as well, we have also chosen AGPL as license for our open source project - https://github.com/patched-codes/patchwork
138.
A comparative study of fine-tuning GPT-4o-mini, Gemini Flash 1.5, Llama-3.1-8B(patched.codes)
2 points by codelion 2y ago|0 comments
139.
by codelion 2y ago
We also tried to implement a GH Issue to PR workflow in patchwork - https://github.com/patched-codes/patchwork It is a bit hard to get it to work reliably except for small changes.
140.
by codelion 2y ago
Yes, we also use treesitter.
141.
Beating GPT-4 on Arena Bench Hard with Patched MOA(patched.codes)
2 points by codelion 2y ago|0 comments
142.
by codelion 2y ago
We have a few large open source projects already using patchwork to help manage issues and PRs - https://github.com/patched-codes/patchwork
143.
by codelion 2y ago
We have a patchflow that does that in patchwork - https://github.com/patched-codes/patchwork it is called ResolveIssue - https://github.com/patched-codes/patchwork/tree/main/patchwo.
144.
by codelion 2y ago
Yeah I was hoping there will be some bit about how LLMs make many of the tasks that people used SpaCy for a lot easier and cheaper. That is a bigger threat to the project going forward I don’t see how it can exist in the same form as today.
145.
Patched RTC: evaluating LLMs for diverse software development tasks(patched.codes)
1 points by codelion 2y ago|0 comments
146.
Working at Heights Safety: Video Analytics Revolution in Construction(securade.ai)
2 points by codelion 2y ago|0 comments
147.
AI in Emergency Response: Improving Situational Awareness and Decision-Making(securade.ai)
1 points by codelion 2y ago|0 comments
148.
Advancing Workplace Safety: AI Video Analytics for Accident Reduction(securade.ai)
2 points by codelion 2y ago|0 comments
149.
by codelion 2y ago
Many such cases. It is very hard to balance composition and abstraction in such frameworks and libraries. And LLMs being so new it has taken several iterations to get the right patterns and architecture while building LLM based apps. With p
150.
How good are LLMs at patching vulnerabilities?(patched.codes)
5 points by codelion 2y ago|0 comments

More ›