PinnedMichał Marcińczuk, Ph.D.inTowards Data ScienceReducing the Size of Docker Images Serving Large Language ModelsHave you encountered a problem where a 1 GB transformer-based model increases even up to 8 GB when deployed using Docker containerization?·6 min read·May 3, 2024--3--3
PinnedMichał Marcińczuk, Ph.D.inCodeNLPCost of running ML experiments on GPU — AWS Cloud vs local GPUDid you know that you are giving away an RTX 4090 for free by running ML experiments for a year?·4 min read·Nov 13, 2023--6--6
PinnedMichał Marcińczuk, Ph.D.inCodeNLPThe training time of the foundation models (from scratch)How many GPU hours does it take to train a large language model (LLM)?·2 min read·Nov 17, 2023--1--1
Michał Marcińczuk, Ph.D.inCodeNLPCross-lingual Named Entity Corpus for Slavic LanguagesWe present a corpus manually annotated with named entities resulting from a series of shared tasks on Named Entity Recognition…·2 min read·8 hours ago----
Michał Marcińczuk, Ph.D.inSeñor PythonDataclasses: an effective use of InitVar in PythonThe story presents how to define init-only properties using the dataclasses library in Python.·4 min read·2 days ago--1--1
Michał Marcińczuk, Ph.D.inTowards Data ScienceReducing the Size of Docker Images Serving Large Language Models (part 2)How to reduce a “small” Docker image by another 10%.·7 min read·May 8, 2024----
Michał Marcińczuk, Ph.D.inCodeNLPKeep an eye on the expenses of your cloud storage while using MLflowMLflow has a tendency to accumulate experiment data, leading to unexpectedly high cloud storage costs. Keep an eye on how much data is on…·5 min read·Mar 27, 2024----
Michał Marcińczuk, Ph.D.inCodeNLPTerminus — a concept of an LLM created in 1968?Terminus is a character in one of the science fiction stories collections named Tales of Pirx the Pilot, which was written by Stanisław…·3 min read·Mar 6, 2024----
Michał Marcińczuk, Ph.D.inCodeNLPUse safetensors to avoid malicious AI modelsEliminate the risk of running malicious AI models by using the write format.·3 min read·Mar 1, 2024----
Michał Marcińczuk, Ph.D.inSeñor PythonA curated list of Python resourcesVerified list of good-quality guidelines, tutorials, stories, and resources related to Python·3 min read·Mar 5, 2024----