Hacker Newsnew | past | comments | ask | show | jobs | submit | kumama's submissionslogin
1.
Token-in-token-out RL training with any agent harness, via a proxy gateway (castform.com)
4 points by kumama 40 days ago | past
2.
Castform now supports supervised finetuning (castform.com)
2 points by kumama 43 days ago | past
3.
how we monitor our rl training runs (castform.com)
1 point by kumama 72 days ago | past
4.
Designing dev onboarding for an agent-first world (castform.com)
2 points by kumama 3 months ago | past
5.
I post-trained a model to reliably roll a die (castform.com)
2 points by kumama 3 months ago | past
6.
Open-Weight Models Don't Need to Win (twitter.com/googrish)
5 points by kumama 4 months ago | past | 8 comments
7.
Prompt caching but for RL – 7.5x speedup on long-prompt/short-response workloads (castform.com)
4 points by kumama 4 months ago | past
8.
Pokegents: Making multi-agent coding feel like a team (castform.com)
8 points by kumama 4 months ago | past | 1 comment
9.
Grpo explained: group relative policy optimization for LLM finetuning (cgft.io)
1 point by kumama 5 months ago | past
10.
Do RL on a model with your vector db (cgft.io)
1 point by kumama 5 months ago | past
11.
What is reinforcement learning finetuning (youtube.com)
3 points by kumama 5 months ago | past
12.
RAG to riches: synthetic data for training RAG agents (cgft.io)
2 points by kumama 6 months ago | past
13.
rag not lag: rl for fast agentic retrieval (cgft.io)
3 points by kumama 6 months ago | past
14.
Show HN: Benchmax, a new open-source RL environment framework for LLM finetuning (github.com/cgftinc)
1 point by kumama on July 29, 2025 | past
15.
Beating o3/o4-mini with Codebase-specific Reinforcement Learning (cgft.io)
3 points by kumama on June 11, 2025 | past
16.
We might be overestimating coding agent performance on SWE-Bench (cgft.io)
1 point by kumama on Nov 5, 2024 | past | 1 comment
17.
How to Improve Code Completion LLMs with Repo-Specific Finetuning (cgft.io)
3 points by kumama on Oct 25, 2024 | past | 1 comment
18.
Show HN: Free AI Code Completion for Xcode with model choice/codebase context (cgft.io)
2 points by kumama on Oct 21, 2024 | past

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: