--- title: Fine-Tuning NVIDIA Nemotron to Know Local Rules, Not... description: NVIDIA shows how to post-train its open Nemotron models with locale-grounded persona datasets so agents follow local rules, not just local language. The... canonical: https://daily.dev/posts/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages-01kseszkv twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Fine-Tuning NVIDIA Nemotron to Know Local Rules, Not Just Local Languages | daily.dev og:description: NVIDIA shows how to post-train its open Nemotron models with locale-grounded persona datasets so agents follow local rules, not just local language. The... og:url: https://daily.dev/posts/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages-01kseszkv og:image: https://api.daily.dev/og/posts/01KseszKV.png og:image:alt: Fine-Tuning NVIDIA Nemotron to Know Local Rules, Not Just Local Languages og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Fine-Tuning NVIDIA Nemotron to Know Local Rules, Not Just Local Languages **[NVIDIA Developer](https://daily.dev/sources/nvidiadev)** · 16 min read · 2 upvotes · 0 comments ## Summary NVIDIA shows how to post-train its open Nemotron models with locale-grounded persona datasets so agents follow local rules, not just local language. The tutorial fine-tunes Nemotron 3.5 Lightning for a French public-service assistant using the Nemotron-Personas-France dataset, built with Pleias as part of a persona data program covering 12 countries. Steps cover defining a behavior contract, generating synthetic persona-based conversations with vLLM, fine-tuning with LoRA via NVIDIA NeMo Automodel, merging the adapter, and evaluating against the base model. In one example, the base model suggests a postal passport application for a resident who can no longer travel, while the fine-tuned model describes a home-visit option requiring a medical certificate less than three months old. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How do I fine-tune NVIDIA Nemotron 3.5 Lightning for a local-language customer service agent? Use the NVIDIA Nemotron repo's workflow: define a behavior contract, generate synthetic persona-based conversations with the sdg/data_designer step using a locale-grounded persona dataset, then fine-tune with LoRA via NVIDIA NeMo Automodel. The example uses nvidia/Nemotron-Personas-France, samples 2,000 personas, serves Nemotron 3.5 Lightning locally with vLLM, and merges the LoRA adapter before evaluating against the base model. _Developers adapting LLMs to local markets can track emerging fine-tuning workflows like this one on daily.dev._ ### Why does the French passport stamp fee differ between mainland France and French Guiana? The passport tax stamp costs €43 for a resident of French Guiana compared to €86 in mainland France, a regional difference that a generic foundation model may not know without locale-specific grounding data such as persona-based training sets tied to official procedure text. _Teams building region-aware AI agents can follow localization fine-tuning approaches like this on daily.dev._ ### What setting do I need for Nemotron 3.5 Lightning to stop it from generating reasoning traces during synthetic data generation? Set enable_thinking: false under providers → extra_body → chat_template_kwargs in the sdg_data_designer.yaml config. Nemotron 3.5 Lightning produces reasoning traces by default; placing this setting under inference_parameters instead causes the run to fail, and omitting it leaves reasoning text in the training data. _Developers debugging LLM config quirks can keep up with practical fixes like this on daily.dev._ ## Similar posts on daily.dev - [How to Fine-Tune Nemotron 3.5 ASR for Your Language, Domain, or Accent](https://daily.dev/posts/how-to-fine-tune-nemotron-3-5-asr-for-your-language-domain-or-accent-lmayxxp44) · Hugging Face · 0 upvotes · 0 comments - [How to Fine-Tune an LLM on NVIDIA GPUs With Unsloth](https://daily.dev/posts/how-to-fine-tune-an-llm-on-nvidia-gpus-with-unsloth-lfqgsiuoc) · NVIDIA · 2 upvotes · 0 comments - [Fine-Tuning NVIDIA Nemotron for Saudi Arabic Dialects, with a Path to Other Languages](https://daily.dev/posts/fine-tuning-nvidia-nemotron-for-saudi-arabic-dialects-with-a-path-to-other-languages-6uqddopfl) · NVIDIA Developer · 0 upvotes · 0 comments - [Fine-tune NVIDIA Nemotron 3 models with Amazon SageMaker AI serverless model customization](https://daily.dev/posts/fine-tune-nvidia-nemotron-3-models-with-amazon-sagemaker-ai-serverless-model-customization-osogwc5ok) · AWS · 1 upvotes · 0 comments - [NVIDIA Nemotron 3.5 Lightning · Ollama Blog](https://daily.dev/posts/nvidia-nemotron-3-5-lightning-ollama-blog-uy1kba1ur) · Ollama · 3 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#deep-learning](https://daily.dev/tags/deep-learning), [#lora](https://daily.dev/tags/lora) [View this post on daily.dev](https://daily.dev/posts/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages-01kseszkv) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Fine-Tuning NVIDIA Nemotron to Know Local Rules, Not Just Local Languages","url":"https://daily.dev/posts/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages-01kseszkv","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages-01kseszkv"},"datePublished":"2026-10-27T16:01:20.720Z","dateModified":"2026-10-27T16:24:29.508Z","description":"NVIDIA shows how to post-train its open Nemotron models with locale-grounded persona datasets so agents follow local rules, not just local language. The...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a6cc2b118c6d8546672297f0e07fd748?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a6cc2b118c6d8546672297f0e07fd748?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"NVIDIA Developer","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"NVIDIA Developer","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/86e45aab42ba48ce83103d01b1119910","url":"https://daily.dev/sources/nvidiadev"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages-01kseszkv","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,deep-learning,lora","timeRequired":"PT16M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"NVIDIA Developer","item":"https://daily.dev/sources/nvidiadev"},{"@type":"ListItem","position":3,"name":"Fine-Tuning NVIDIA Nemotron to Know Local Rules, Not Just Local Languages"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages-01kseszkv#faq","mainEntity":[{"@type":"Question","name":"How do I fine-tune NVIDIA Nemotron 3.5 Lightning for a local-language customer service agent?","acceptedAnswer":{"@type":"Answer","text":"Use the NVIDIA Nemotron repo's workflow: define a behavior contract, generate synthetic persona-based conversations with the sdg/data_designer step using a locale-grounded persona dataset, then fine-tune with LoRA via NVIDIA NeMo Automodel. The example uses nvidia/Nemotron-Personas-France, samples 2,000 personas, serves Nemotron 3.5 Lightning locally with vLLM, and merges the LoRA adapter before evaluating against the base model. Developers adapting LLMs to local markets can track emerging fine-tuning workflows like this one on daily.dev."}},{"@type":"Question","name":"Why does the French passport stamp fee differ between mainland France and French Guiana?","acceptedAnswer":{"@type":"Answer","text":"The passport tax stamp costs €43 for a resident of French Guiana compared to €86 in mainland France, a regional difference that a generic foundation model may not know without locale-specific grounding data such as persona-based training sets tied to official procedure text. Teams building region-aware AI agents can follow localization fine-tuning approaches like this on daily.dev."}},{"@type":"Question","name":"What setting do I need for Nemotron 3.5 Lightning to stop it from generating reasoning traces during synthetic data generation?","acceptedAnswer":{"@type":"Answer","text":"Set enable_thinking: false under providers → extra_body → chat_template_kwargs in the sdg_data_designer.yaml config. Nemotron 3.5 Lightning produces reasoning traces by default; placing this setting under inference_parameters instead causes the run to fail, and omitting it leaves reasoning text in the training data. Developers debugging LLM config quirks can keep up with practical fixes like this on daily.dev."}}]} ```