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

• URL: http://arxiv.org/abs/2609.01572v1
• Date: Tue, 01 Sep 2026 17:39:26 GMT
• Title: From Production Traffic to Post-Training: Building a Self-Hosted LLM That Covers the Corporate Request Mix
• Abstract summary: We consolidate traffic from over 200 internal applications onto a single model.<n>We train a separate GRPO expert per axis and merge them via two-stage SLERP.<n>The model absorbs 50% of platform traffic, 116M requests per month, at a fraction of the serving cost.
• Score: 55.08143827481755
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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 optimising all objectives jointly, which introduces cross-domain reward interference, we train a separate GRPO expert per axis and merge them via two-stage SLERP. Each expert's reward exposes a distinct failure mode, namely semantic collapse, over-calling, and verbosity hacking, each requiring a domain-specific fix. In non-reasoning mode the recipe surpasses a ${\sim}7\times$ larger by total parameters baseline on the in-house Arena with 69.6 to 65.8, instruction following with 0.85 to 0.83, and function-calling with 0.79 to 0.77, while lifting general dialogue benchmarks. The model absorbs 50% of platform traffic, 116M requests per month, at a fraction of the serving cost.

Related papers

• ClawArena-Team: Benchmarking Subagent Orchestration and Dynamic Workflows in Language-Model Agents [61.184954205350984]
  ClawArena-Team is a benchmark of 41 multi-turn, multimodal, multi-agent scenarios spanning 258 evaluation rounds and 72 staged updates.<n>All scoring is execution-based with no LLM judge: an overall score -- the Subagent-Management Score (SMS) -- multiplies task correctness by a least-privilege and modality-routing factor.
  arXiv  Detail & Related papers  (2026-06-30T06:06:08Z)
• Sub-Billion, Super-Frontier: Small Language Models Rival Zero-Shot Frontier LLMs on General and Literary Relation Extraction [5.070542698701158]
  Large language models (LLMs) achieve strong relation extraction (RE)<n>We investigate how far small language models (SLMs) can close this gap across general-domain and literary text.
  arXiv  Detail & Related papers  (2026-06-21T17:24:31Z)
• Trace2Policy: From Expert Behavior Traces to Self-Evolving Decision Agents [5.689042186242701]
  Decision rules that enterprise experts apply tacitly can be systematically recovered and improved through iterative error analysis.<n>We present textbfTrace2Policy, whose core mechanism is textbfEISR.<n>Each round executes the rules on a validation set, clusters errors by root cause into MISSING, WRONG, or CONFLICT types, applies targeted patches, and commits only those that pass a regression gate.
  arXiv  Detail & Related papers  (2026-06-09T06:05:29Z)
• ChipMATE: Multi-Agent Training via Reinforcement Learning for Enhanced RTL Generation [55.947962672433675]
  ChipMATE is the first self-trained multi-agent framework for RTL generation.<n>Inspired by industrial practice, ChipMATE pairs a Verilog agent with a Python reference-model agent that mutually verify each other's outputs.<n>ChipMATE achieves 75.0% and 80.1% pass@1 on VerilogEval V2 with 4B and 9B base models.
  arXiv  Detail & Related papers  (2026-05-13T01:04:21Z)
• CAX-Agent: A Lightweight Agent Harness for Reliable APDL Automation [7.355373109826612]
  This paper presents the architecture of CAX-Agent, a lightweight agent harness for MAPDL automation.<n>We evaluate three recovery strategies (no_recovery, rule_only, and model_only) on 50 standard structural benchmarks.<n>Model_only achieves the best completion rate (0.9267), task score (3.59/4), total score (9.16/10), and zero-intervention rate (0.84), outperforming rule_only (0.7733, 3.17/4, 7.03/10, 0.00) and no_recovery (0.6933, 2.74/4, 5.60/10, 0.00) with large effect sizes
  arXiv  Detail & Related papers  (2026-05-12T14:46:34Z)
• ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox [61.862814740220806]
  $textbfComplexMCP$ is a benchmark designed to evaluate agents in rigorous conditions.<n>Built on the Model Context Protocol (MCP), $textbfComplexMCP$ provides over 300 meticulously tested tools.
  arXiv  Detail & Related papers  (2026-05-11T16:20:51Z)
• Chimera: Latency- and Performance-Aware Multi-agent Serving for Heterogeneous LLMs [62.17306142810532]
  We present Chimera, a predictive scheduling system for multi-agent workflow serving on heterogeneous LLM clusters.<n>Chime traces the best latency, reducing end-to-end latency by 1.2--2.4$times$ and improving task performance by 8.0-9.5 percentage points on average over competitive baselines.
  arXiv  Detail & Related papers  (2026-03-23T17:01:42Z)
• Efficient Multi-Model Orchestration for Self-Hosted Large Language Models [2.3275796286410677]
  Pick and Spin is a framework that makes self-hosted orchestration and economical.<n>It integrates a unified Helm-based deployment system, adaptive scale-to-zero automation, and a hybrid routing module.<n>It achieves up to 21.6% higher success rates, 30% lower latency, and 33% lower cost per query compared with static deployments of the same models.
  arXiv  Detail & Related papers  (2025-12-26T22:42:40Z)
• Evaluating the Limitations of Local LLMs in Solving Complex Programming Challenges [0.31498833540989407]
  This study examines the performance of open-source, locally hosted large-language models (LLMs) in handling complex programming tasks.<n>Building on the original Framework for AI-driven Code Generation Evaluation (FACE), the authors retrofit the pipeline to work entirely offline.<n>Results show that the overall pass@1 accuracy is modest for the local models, with the best models performing at approximately half the acceptance rate of the proprietary models.
  arXiv  Detail & Related papers  (2025-09-18T14:13:30Z)
• Towards Efficient Automatic Self-Pruning of Large Language Models [55.90119819642064]
  Post-training structured pruning is a promising solution that prunes Large Language Models without the need for retraining.<n>We argue that the key to mitigating this issue lies in accurately determining the pruning rate for each layer.<n>We introduce $textbfSelf-Pruner$ an end-to-end automatic self-pruning framework for LLMs, which efficiently search layer-wise pruning rates.
  arXiv  Detail & Related papers  (2025-02-20T09:59:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.