from Hacker News

robertnishihara

joined 10/16/17, 6:03 AM has 366 karma

• Roll: Reinforcement Learning Optimization for Large-Scale Learning

  by robertnishihara on 6/14/25, 5:40 AM, with 0 comments

• An Open Source Stack for AI Compute: Kubernetes and Ray and PyTorch and VLLM

  by robertnishihara on 6/13/25, 4:40 PM, with 0 comments

• Uv and Ray: Pain-Free Python Dependencies in Clusters

  by robertnishihara on 2/28/25, 4:28 AM, with 0 comments

• Ray Batch Inference at Pinterest (Part 3)

  by robertnishihara on 10/16/24, 7:51 AM, with 0 comments

• Direct Preference Optimization with Synthetic Data on Anyscale

  by robertnishihara on 8/21/24, 4:29 PM, with 0 comments

• Building an LLM Router for High-Quality and Cost-Effective Responses

  by robertnishihara on 7/2/24, 2:57 AM, with 0 comments

• Ray Infrastructure at Pinterest

  by robertnishihara on 6/18/24, 5:09 AM, with 0 comments

• Lessons from training a Stable Diffusion model on 2B images

  by robertnishihara on 5/11/24, 12:15 AM, with 0 comments

• Canva Built a Modern AI Platform Using Anyscale

  by robertnishihara on 4/3/24, 4:47 PM, with 0 comments

• Building RAG-Based LLM Applications for Production

  by robertnishihara on 2/14/24, 9:49 PM, with 0 comments

• Fine-tuning LLMs for longer context and better RAG systems

  by robertnishihara on 2/13/24, 10:51 PM, with 0 comments

• Two-day hands-on RAG Bootcamp for developers

  by robertnishihara on 1/31/24, 7:25 PM, with 0 comments

• RAG at Scale: 10x Cheaper Embedding Computations with Anyscale and Pinecone

  by robertnishihara on 1/16/24, 8:41 PM, with 0 comments

• Comparing LLM Performance: Introducing the Open Source Leaderboard for LLM APIs

  by robertnishihara on 12/21/23, 8:17 PM, with 0 comments

• LLMPerf Leaderboard

  by robertnishihara on 12/21/23, 6:43 PM, with 0 comments

• Anyscale Endpoints: JSON Mode and Function Calling Features

  by robertnishihara on 12/14/23, 4:52 AM, with 0 comments

• LLM summarization: A case study of human, Llama-2, & GPT-4 summarization quality

  by robertnishihara on 11/10/23, 10:25 PM, with 0 comments

• Reproducible Performance Metrics for LLM Inference

  by robertnishihara on 11/2/23, 6:00 AM, with 0 comments

• Building Rag-Based LLM Applications for Production

  by robertnishihara on 10/25/23, 10:54 PM, with 0 comments

• Anyscale Endpoints: LLM inference and fine-tuning

  by robertnishihara on 10/25/23, 12:31 AM, with 0 comments

• Anyscale Private Endpoints and Anyscale Endpoints Fine-Tuning

  by robertnishihara on 10/24/23, 11:59 PM, with 0 comments

• Loading Llama-2 70B 20x faster with Anyscale Endpoints

  by robertnishihara on 10/12/23, 3:11 AM, with 0 comments

• A Comprehensive Guide for Building Rag-Based LLM Applications

  by robertnishihara on 9/14/23, 6:33 AM, with 48 comments

• Llama 2 is about as factually accurate as GPT-4 for summaries and is 30X cheaper

  by robertnishihara on 8/23/23, 5:26 PM, with 0 comments

• Nearly all LLMs will be multi-modal

  by robertnishihara on 8/15/23, 7:34 PM, with 1 comments

• ByteDance Scales Offline Inference with Multi-Modal LLMs to 200 TB Data

  by robertnishihara on 8/15/23, 5:23 PM, with 0 comments

• Fine-Tuning Llama-2: A Comprehensive Case Study for Tailoring Custom Models

  by robertnishihara on 8/11/23, 4:34 PM, with 59 comments

• Distributed Machine Learning at Instacart

  by robertnishihara on 3/19/23, 4:10 AM, with 0 comments