kumama

10 ms·

31.
by kumama 5mo ago
yeah :)
32.
Open-Weight Models Don't Need to Win(twitter.com)
5 points by kumama 5mo ago|8 comments
33.
Prompt caching but for RL – 7.5x speedup on long-prompt/short-response workloads(castform.com)
4 points by kumama 5mo ago|0 comments
34.
Pokegents: Making multi-agent coding feel like a team(castform.com)
8 points by kumama 5mo ago|1 comments
35.
Grpo explained: group relative policy optimization for LLM finetuning(cgft.io)
1 points by kumama 6mo ago|0 comments
36.
Do RL on a model with your vector db(cgft.io)
1 points by kumama 6mo ago|0 comments
37.
What is reinforcement learning finetuning(youtube.com)
3 points by kumama 6mo ago|0 comments
38.
RAG to riches: synthetic data for training RAG agents(cgft.io)
2 points by kumama 7mo ago|0 comments
39.
rag not lag: rl for fast agentic retrieval(cgft.io)
3 points by kumama 7mo ago|0 comments
40.
Show HN: Benchmax, a new open-source RL environment framework for LLM finetuning(github.com)
1 points by kumama 1y ago|0 comments
41.
Beating o3/o4-mini with Codebase-specific Reinforcement Learning(cgft.io)
3 points by kumama 1y ago|0 comments
42.
We might be overestimating coding agent performance on SWE-Bench(cgft.io)
1 points by kumama 2y ago|1 comments
43.
by kumama 2y ago
Hey everyone! We recently came across a ICLR submission highlighting dataset contamination issues with SWE-Bench. After filtering out those issues, the authors saw the performance of SWE-Agent + GPT-4 drop significantly, from 12.47% to 3.97
44.
How to Improve Code Completion LLMs with Repo-Specific Finetuning(cgft.io)
3 points by kumama 2y ago|1 comments
45.
by kumama 2y ago
Hey everyone! We've been working on helping eng teams finetune custom code LLMs for their specific internal code repos for different tasks across the SDLC. We wrote a blog post about how we're doing it for code completions. We ess
46.
Show HN: Free AI Code Completion for Xcode with model choice/codebase context(cgft.io)
2 points by kumama 2y ago|0 comments
47.
by kumama 3y ago
cool tool