Igor Gitman – NVIDIA Technical Blog News and tutorials for developers, data scientists, and IT admins 2025-07-10T18:30:06Z http://www.open-lab.net/blog/feed/ Igor Gitman <![CDATA[How to Streamline Complex LLM Workflows Using NVIDIA NeMo-Skills]]> http://www.open-lab.net/blog/?p=102597 2025-07-10T18:30:06Z 2025-06-25T17:13:59Z A typical recipe for improving LLMs involves multiple stages: synthetic data generation (SDG), model training through supervised fine-tuning (SFT) or...]]>

A typical recipe for improving LLMs involves multiple stages: synthetic data generation (SDG), model training through supervised fine-tuning (SFT) or reinforcement learning (RL), and model evaluation. Each stage requires using different libraries, which are often challenging to set up and difficult to use together. For example, you might use NVIDIA TensorRT-LLM or vLLM for SDG and NVIDIA…

Source

]]>
Igor Gitman <![CDATA[Build Custom Reasoning Models with Advanced, Open Post-Training Datasets]]> http://www.open-lab.net/blog/?p=98680 2025-05-29T19:05:03Z 2025-05-14T16:33:26Z Synthetic data has become a standard part of large language model (LLM) post-training procedures. Using a large number of synthetically generated examples from...]]>

Synthetic data has become a standard part of large language model (LLM) post-training procedures. Using a large number of synthetically generated examples from either a single or cohort of open-source, commercially permissible LLMs, a base LLM is finetuned either with supervised finetuning or RLHF to gain instruction-following and reasoning skills. This process can be seen as a knowledge…

Source

]]>
Igor Gitman <![CDATA[Mixed Precision Training for NLP and Speech Recognition with OpenSeq2Seq]]> http://www.open-lab.net/blog/?p=12300 2022-08-21T23:39:09Z 2018-10-09T13:00:45Z The success of neural networks thus far has been built on bigger datasets, better theoretical models, and reduced training time. Sequential models, in...]]>

Source

]]>
1
���˳���97caoporen����