Loading…
Topic
622 posts about Machine Learning. Every summary links to the original.
See the idea
A walkthrough of how a 3D CT model turns a voxel volume into a structured diagnostic report: isotropic resampling, 3D patch tokens, spatial encoding, and language-model reasoning.
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
Newest first
Transformer-based automatic speech recognition models like Wav2Vec2 crash on long audio inputs because the O(n²) attention mechanism rapidly exhausts GPU memory. Simple audio chunking avoids out-of-memory errors but causes poor transcription accuracy at chunk boundaries due to a lack of surrounding context. To resolve this, Hugging Face Transformers leverages the Connectionist Temporal Classification architecture to process overlapping audio chunks with configurable strides. The pipeline discards low-quality logits at chunk edges and chains the remaining central predictions together to reconstruct a seamless transcript. This striding technique operates out of the box with language model-augmented pipelines and adapts directly to low-latency live audio streaming.
Programmatically searching the Hugging Face Hub previously required navigating web browser widgets or enduring trial and error to guess exact query string formats. The huggingface_hub library solves this issue by introducing helper utilities such as ModelSearchArguments, DatasetSearchArguments, and ModelFilter alongside HfApi. These namespace helpers translate accessible Python attributes into the formatted parameters expected by the backend API, covering datasets, tasks, and libraries. For complex queries across multiple tasks, frameworks, and datasets, developers can combine criteria inside ModelFilter instances and retrieve matching model metadata via api.list_models. Under the hood, the library uses AttributeDictionary, a data structure inspired by fastcore that enables nested tab-completion while supporting standard dictionary key indexing for special characters.
Hugging Face announced an official integration with Stable-Baselines3, a popular PyTorch library for training and testing Deep Reinforcement Learning agents across diverse environments like Gym, Atari, MuJoco, and Procgen. The integration enables researchers and developers to host their saved reinforcement learning checkpoints on the Hugging Face Hub and download pre-trained community models. Interacting with the Hub requires installing the huggingface_hub and huggingface_sb3 packages, which supply helper methods for authentication, downloading, and uploading. Practitioners can retrieve checkpoint zip files using the load_from_hub function by providing the target repository identifier and filename before loading them into Stable-Baselines3 algorithms. Furthermore, users authenticated via CLI or notebook login can train policies such as PPO and publish their saved zip files to the Hub using push_to_hub.
Search at Shopify is presented as a case for range across data science and engineering, rather than narrow specialization in either discipline. The article describes a team that treats work as “search” instead of separating data and engineering responsibilities: engineers learn experimentation and model evaluation, while data scientists build performant, testable, maintainable production systems. It frames search as a sequence of trade-offs among relevance, correctness, scalability, performance, stability, and maintainability, arguing that people who understand only half the constraints can produce siloed handoffs, deployment delays, or suboptimal systems. Shopify’s stated model is a single collaborative team with shared planning and execution, intended to reduce territorial boundaries and support smart, fast, scalable search. The article extends this principle to personalized and conversational products, concluding that future work will require practitioners to move between data and engineering.
Deploying large Transformer models at scale often faces severe prediction latency bottlenecks, driving up infrastructure costs and limiting real-time production use cases. To address this challenge, Hugging Face evaluated Infinity, a containerized hardware-optimized inference solution paired with the Infinity Multiverse model optimization service. Testing covered 192 configurations on Amazon EC2 C6i instances powered by 3rd generation Intel Xeon Scalable processors across varying CPU cores, sequence lengths, and batch sizes. The benchmarks demonstrated that an Ice Lake-optimized DistilBERT container achieved up to 800% higher throughput than vanilla Transformers and delivered 1 to 4 millisecond end-to-end latencies for sequence lengths up to 64 tokens. Although Infinity was later discontinued in favor of Inference Endpoints and Optimum libraries, the results demonstrated substantial efficiency gains on modern CPU hardware.
Wav2Vec2 models fine-tuned with Connectionist Temporal Classification transcribe speech without external language models, but decoding can still suffer from spelling inaccuracies. Hugging Face Transformers addressed this by integrating Kensho Technologies' pyctcdecode library to support decoding with n-gram language models. Instead of decoding using simple argmax operations over logits, the Wav2Vec2ProcessorWithLM class feeds full probability matrices into beam search guided by KenLM n-gram probabilities. KenLM's build_binary utility compresses language model ARPA files into binary formats, reducing file size by more than half for faster loading and hub deployment. In Swedish xls-r-300m-sv benchmarks on Common Voice 7, this 5-gram boosted decoding setup achieved an 18.85% word error rate, delivering an approximate 30% relative performance gain.
Deploying EleutherAI's 6 billion parameter GPT-J model for production inference presents latency hurdles due to large memory footprints and lengthy startup times. Loading the model via standard methods takes up to several minutes, conflicting with strict real-time response limits such as Amazon SageMaker's 60-second threshold. To overcome this limitation, the model is serialized using PyTorch's native save mechanisms, packaged into a compressed archive with supporting assets, and stored on Amazon S3. This alternative loading workflow reduces GPT-J load times down to 7.7 seconds. An Amazon SageMaker real-time endpoint is then deployed on an NVIDIA T4 GPU instance using the Hugging Face Inference Toolkit.
Active learning requires iteratively adding labeled data, retraining models, and serving them to end users. Building such pipelines often demands substantial effort in data labeling, model selection, hyperparameter tuning, and training infrastructure. The author demonstrates a low-code active learning pipeline using Explosion's Prodigy for entity annotation and Hugging Face's AutoNLP for automatic training and evaluation. After first training a news categorization model achieving 98.67% accuracy on Kaggle's BBC News dataset, the author manually annotated named entities across iterative batches. An export script converted annotations to JSONL with IOB tags, showing progressive metric improvements from 20 samples to 250 samples, where the token classification model attained 95.9% accuracy, 0.73 precision, and 0.79 recall.
Shopify presents a five-level Data Science Hierarchy of Needs that organizes data work by the value it delivers, rather than by technical sophistication. Its tool-agnostic philosophy favors simple solutions first, then more complex analysis when necessary, with impact and better decisions as the goal. The model begins with collecting and modeling clean, conformed data using dimensional modeling, then moves through description, prediction and inference, prescription, and influence. Shopify applied this structure to a COVID-19 impact analysis, using statistical analysis to examine regional effects and inform measures including an extended free trial, expanded funding, shipping options, and local commerce features. The account says the analysis used neither artificial intelligence nor machine learning, yet supported decisions for merchants and Shopify during the pandemic.
Hugging Face has acquired Gradio, the open-source machine learning library designed for building and sharing interactive model interfaces. Gradio originated in 2019 when its founder struggled to share a medical computer vision model with a physician collaborator who did not write Python. Co-founded alongside Ali Abdalla, Ali Abid, and Dawood Khan, the project expanded from computer vision into text, speech, and video modalities. Over 300,000 demos have been built using Gradio, allowing interdisciplinary industry teams and researchers to debug models internally and showcase them externally to non-technical users. The acquisition unites Gradio with Hugging Face to broaden browser-based machine learning accessibility and expand hiring efforts across the joint team.
Standard Transformer architectures scale poorly in compute and memory because pairwise dot-product self-attention depends quadratically on input size. Perceiver IO addresses this constraint by computing self-attention across a small set of latent variables rather than directly on high-dimensional inputs. Inputs and outputs interact with the model via cross-attention operations, decoupling compute and memory costs from input and output dimensions. Integrated into Hugging Face Transformers via the PerceiverModel class, the architecture supports diverse data types using optional preprocessors, decoders, and postprocessors. Experiments demonstrate competitive performance across text, multimodal video classification, 3D point cloud classification on ModelNet40, and StarCraft II reinforcement learning in AlphaStar.
Training code generation models from scratch requires careful data filtering, tokenization, and compute management. To build CodeParrot, a 1.5-billion-parameter Python generation model, engineers extracted 20 million Python files from BigQuery and cleaned the resulting 180 GB corpus down to 50 GB after discovering extreme duplication. A custom GPT-2 tokenizer was trained over streamed samples, and the GPT-2 large architecture was initialized with layer-scaled, full-precision attention mechanisms. The training pipeline leveraged Hugging Face Accelerate alongside gradient checkpointing and a streaming iterable dataset designed to yield fixed-length concatenated token sequences. CodeParrot was trained on roughly 25 to 30 billion tokens and evaluated on coding tasks using the pass@k metric.
Hugging Face has released Snowball Fight 1vs1, its first custom Deep Reinforcement Learning environment created using the Unity Machine Learning Agents Toolkit. The game enables players to shoot snowballs against a trained agent and is hosted on Hugging Face Spaces. This release marks the beginning of an effort to build an ecosystem for Deep Reinforcement Learning researchers and enthusiasts using Unity ML-Agents. The initiative focuses on building open-source environments, hosting models and training environments on the Hugging Face Hub, and deploying interactive demos via Spaces. Future plans include publishing technical tutorials, releasing additional environments, and developing a Snowball Fight 2vs2 version that utilizes the MA-POCA algorithm to train cooperative team behaviors.
Transformer prediction speed can hinder latency-sensitive workloads like conversational systems and search, while optimizing hardware performance requires significant effort. Hugging Face introduced Optimum, an open-source library that reduces Transformer model latency across diverse hardware targets. Through a partnership with Graphcore, BERT is introduced as the initial IPU-optimized model for parallel AI hardware. Developers configure the Poplar SDK environment and PopTorch inside Graphcloud, install optimum-graphcore, and execute the run_qa.py script with an IPU configuration file. Fine-tuning bert-base-uncased on SQuAD v1.1 using the IPUTrainer class yields an evaluation F1 score of 88.2757 and an exact match score of 80.6623 across 10,784 evaluation samples.
Hugging Face released an open-source Python library and no-code interface named the Data Measurements Tool to calculate metrics for responsible dataset curation. In AI development, scraped data often enters training pipelines without detailed analysis of representation or downstream model influence, partly because querying raw datasets typically requires complex coding. Built using Hugging Face Dataset and Spaces Hubs alongside Streamlit, the tool provides descriptive, distributional, and comparison statistics. It computes missing values, vocabulary sizes, duplicate counts, label distributions, and Zipf's law adherence to identify unnatural language patterns or markup artifacts. Additionally, it offers embedding clusters via a Sentence-Transformer model and normalized pointwise mutual information metrics across gender and sexual orientation identity terms.
Transfer learning makes CPU-based clusters a viable, cost-effective option for shorter deep learning fine-tuning jobs. To accelerate PyTorch workloads, clusters can leverage Intel Ice Lake architecture featuring AVX-512 and Vector Neural Network Instructions alongside specialized software optimizations. Implementing the Intel Extension for PyTorch and the oneAPI Collective Communications Library (oneCCL) addresses communication bottlenecks across distributed nodes. In demonstrated tests fine-tuning transformer models on GLUE tasks across Amazon EC2 c6i.16xlarge instances, two-node jobs delivered roughly a 1.7x speedup, while four-node setups achieved a 3x speedup. These performance libraries and multi-node configurations maintain consistent scaling across both smaller datasets like MRPC and larger benchmarks like QQP.
Fine-tuning the multilingual XLS-R speech model adapts pretrained cross-lingual audio representations to automatic speech recognition tasks with limited labeled data. The demonstration configures a Wav2Vec2-XLS-R-300M checkpoint using Connectionist Temporal Classification on roughly four hours of validated Turkish audio from Common Voice. Building the pipeline requires pairing a Wav2Vec2FeatureExtractor for audio signals with a custom Wav2Vec2CTCTokenizer derived from dataset transcriptions. Training with Hugging Face Trainer over 3,200 steps decreases the validation word error rate from 0.7000 down to 0.3195. While the resulting transcription demonstrates recognizable phonetic alignment, output quality can be further improved by extending training schedules, refining preprocessing, and adding language model decoding.
Shopify’s data team sought to estimate whether accepting Shopify Capital for the first time increased merchants’ subsequent sales, without randomly denying funding in an A/B test. They compared US first-time adopters in January 2019 with Canadian shops that later pre-qualified for Capital, using propensity scores built from Shopify sales and performance characteristics and matching them with a caliper-based greedy nearest-neighbor algorithm. After matching about 600 treated shops against roughly 8,000 controls, they evaluated log-transformed cumulative GMV from February through July 2019 with a binary regression and checked balance using standardized mean differences, visual comparisons, and variance ratios. The estimated geometric-average GMV was 36% higher for US merchants, with a 95% confidence interval of 13% to 65%; robustness checks supported a positive treatment effect.
Scaling transformer inference on modern CPUs requires coordinating hardware capabilities with software optimizations across memory allocation, parallel execution, and mathematical kernel libraries. Intel Ice Lake Xeon processors deliver up to seventy-five percent faster inference across natural language processing tasks compared to Cascade Lake by combining Sunny Cove architecture improvements with dedicated extensions like Intel Extension for PyTorch. Fine-tuning software knobs such as alternative allocators, threading runtimes like Intel OpenMP, and core allocations can involve thousands of possible configuration combinations. Rather than evaluating these combinations through exhaustive brute force, Bayesian optimization via Intel SigOpt achieves comparable latency within roughly twenty trials and reveals experiment-dependent parameter importance. These software tuning strategies enable low-latency production transformer serving without dedicated accelerators.
Hugging Face scheduled the release of Part 2 of its machine learning course for November 15th alongside a collaborative community event. While Part 1 covered text classification and Model Hub uploads, Part 2 expands into token classification, causal and masked language modeling, translation, summarization, question answering, and ecosystem tools like Datasets and Tokenizers. The event includes two days of technical talks from industry practitioners covering transfer learning, ethics in machine learning development, and workflows across Keras, PyTorch, and TensorFlow. Participants collaborate on team projects to fine-tune natural language processing models and develop interactive portfolio demos, with AWS supplying complimentary compute through Amazon SageMaker. Attendees who successfully complete and demonstrate a project receive a certificate of completion.
The best engineering reads, once a week.
