From Production Traffic to Post-Training: Building a Self-Hosted LLM That Covers the Corporate Request Mix

• 类型：arxiv
• 标识：2609.01572
• 链接：https://arxiv.org/abs/2609.01572
• 主分类：engineering
• 形态：benchmark
• TLDR：Data-residency constraints force enterprises to self-host LLMs, but continuous adoption of newer models without decommissioning their predecessors expands the serving fleet, fragmenting a finite GPU pool. We consolidate traffic from over 200 internal applications onto a single model by closing quality gaps identified through production error analysis along three axes: instruction following, function-calling, and internal task distribution. Quality is tracked by offline benchmarks stratified to production traffic and scored by deterministic verifiers or calibrated LLM judges. Rather than optimi
• 待LLM分类：否
• 标题中文：从生产流量到后训练：构建覆盖企业请求组合的自托管 LLM
• TLDR中文：数据驻留约束迫使企业自托管 LLM，但不断引入新模型而不下线旧模型会扩张服务集群，分散有限的 GPU 池。我们通过沿指令遵循、函数调用和内部任务分布三个维度，针对生产错误分析所发现的质量差距，将 200 多个内部应用的流量整合到单一模型上。质量通过按生产流量分层的离线基准进行跟踪，并由确定性验证器或经过校准的 LLM 评判器打分。不同于针对
• 来源文件：
• /inbox/tom/_candidates/2026-09-02-agent-rag-longcontext-candidates.json