<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Amardeep Kumar</title><description>ML Engineer &amp; NLP Researcher. Building at the intersection of LLMs, research, and production AI systems.</description><link>https://ad6398.github.io/</link><item><title>Fine-tuning Phi-2 with DPO on the Anthropic HH Dataset</title><link>https://ad6398.github.io/posts/rlhf-phi2-dpo/</link><guid isPermaLink="true">https://ad6398.github.io/posts/rlhf-phi2-dpo/</guid><description>Fine-tuning Microsoft&apos;s Phi-2 using Direct Preference Optimization (DPO) on the Anthropic Helpful and Harmless dataset with LoRA and 8-bit quantization.</description><pubDate>Fri, 01 Mar 2024 00:00:00 GMT</pubDate></item><item><title>How We Cut ML Inference Latency by 40% on Kubernetes</title><link>https://ad6398.github.io/posts/async-model-serving-40-percent/</link><guid isPermaLink="true">https://ad6398.github.io/posts/async-model-serving-40-percent/</guid><description>The architecture behind our async model serving platform at Instabase — async workers, RabbitMQ, multi-level caching, and sticky routing to cut inference time by 40%.</description><pubDate>Thu, 01 Jun 2023 00:00:00 GMT</pubDate></item><item><title>GupShup: Summarizing Code-Switched Conversations</title><link>https://ad6398.github.io/posts/gupshup-code-switching-summarization/</link><guid isPermaLink="true">https://ad6398.github.io/posts/gupshup-code-switching-summarization/</guid><description>Our EMNLP 2021 paper on abstractive summarization of Hindi-English code-switched conversations — introducing the GupShup dataset.</description><pubDate>Sun, 07 Nov 2021 00:00:00 GMT</pubDate></item></channel></rss>