Hacker News: aseligmanhttps://news.ycombinator.com/user?id=aseligmanHacker News RSShttps://hnrss.org/hnrss v2.1.1Sun, 03 May 2026 03:46:08 +0000<![CDATA[New comment by aseligman in "Lossless LLM compression for efficient GPU inference via dynamic-length float"]]>Some additional context: many real world agent use cases struggle to balance quality, cost, and performance. This technique can help avoid the tradeoffs that quantization techniques introduce, including unpredictable results while you try cost optimize an agent. In some cases the cost savings can be significant using dfloat11 as you squeeze into more affordable GPUs.

* I work with xmad.ai

]]>Fri, 25 Apr 2025 22:06:53 +0000https://news.ycombinator.com/item?id=43798970aseligmanhttps://news.ycombinator.com/item?id=43798970https://news.ycombinator.com/item?id=43798970<![CDATA[New comment by aseligman in "Salesforce Wear Developer Pack"]]>We learned a lot: https://developer.salesforce.com/blogs/developer-relations/2...

]]>Sun, 15 Jun 2014 03:47:23 +0000https://news.ycombinator.com/item?id=7894314aseligmanhttps://news.ycombinator.com/item?id=7894314https://news.ycombinator.com/item?id=7894314