Models

An open-source pipeline for training natural language understanding models for text classification and named entity recognition, which automatically chooses between few-shot and full fine-tuning methods based on how much labeled data is available.

• Models

A fine-tuning toolkit for adapting general-purpose language models to specialized domains, such as legal, medical or financial text, that analyzes a dataset's token distribution and class balance to propose a fine-tuning strategy while still allowing full manual control.

• Models

A multi-task learning framework built on the modernBERT architecture that trains several NLP tasks together in one shared network, with dynamic loss-weighting and task scheduling intended to reduce GPU hours compared with fine-tuning a separate model per task.

• Models

A callback library for Hugging Face's Trainer that forwards training events, such as loss and epoch boundaries, to an observability backend through Pydantic Logfire, without a training script needing to be modified or a custom callback subclass written for each project.

• Models
• Monitoring

A progressive, curriculum-style collection of scripts and notebooks for fine-tuning Hugging Face Transformers models, moving from the high-level Trainer API down to a hand-written training loop, tokenizer details, and methods such as LoRA and prefix tuning.

• Models
• Tutorials

A configuration-driven fine-tuning framework for Hugging Face Diffusers pipelines, such as Stable Diffusion, where a YAML file describes the whole training pipeline, including which layers to freeze or train, instead of a hand-written training script.

• Models
• Generative media

A research framework, released with its NeurIPS 2026 paper, that pairs a vision-language reasoning model with a JEPA-based world model, so high-level semantic reasoning guides low-level physical prediction for embodied, robotic tasks.

• Models
• Papers

A family of large language models trained entirely on domestic Chinese computing infrastructure by China Telecom's AI research institute, released with reasoning-focused variants.

• Models

A modular framework for post-training large language models, providing unified GPU backends and system optimizations for reinforcement learning that plug into existing training frameworks.

• Models

A PyTorch runtime for full-parameter fine-tuning of large language models on a single GPU, which streams inactive layers between GPU memory, CPU RAM and optional NVMe storage.

• Models

A multi-framework neural-network compiler that lowers PyTorch, JAX, Flax and Keras models into portable GGUF binaries and standalone, human-readable C++ code with CPU and CUDA GPU execution.

• Models
• Inference

A SwiftUI workbench for discovering, converting, running and benchmarking machine-learning models with Apple's CoreAI framework, covering language, vision, audio and diffusion model recipes.

• Models
• Inference

A model-training and code-synthesis toolchain for Analog Devices' MAX78000 and MAX78002 ultra-low-power machine-learning accelerator chips, converting a trained PyTorch model into C code for the device.

• Models
• Inference

Inference and embedding code for LucaOne, a foundation model for nucleic-acid and protein sequences, with Hugging Face support for fine-tuning on sequence- and residue-level biological tasks.

• Models

An operating layer for a pair of NVIDIA DGX Spark machines that lets several large models share the hardware as switchable lanes, plus on-demand model loading and QLoRA fine-tuning on the idle node.

• Models
• Inference

A reference training pipeline that grows a single architecture from a 2B dense seed to a 120B sparse mixture-of-experts model through four stages on a single eight-GPU node.

• Models

A curated, actively maintained list of open-weight AI models selected for commercially exploitable licenses, no geographic restrictions, and verified benchmarks, organized by category and VRAM tier.

• Models
• Curated lists

A native macOS app that puts an entire small-language-model training workflow, from architecture to tokenizer to fine-tuning to chat, in one place, using Apple Silicon and MLX.

• Models

An interactive canvas for inspecting a large language model's weights tensor by tensor, showing measured INT8, INT4 and FP8 quantization error, spectra and outlier channels for every tensor in a checkpoint.

• Models

A Rust command-line tool that filters large synthetic and pretraining text datasets by sending each row through TypeSafe AI's Jev model for calibrated, rubric-based judgments instead of a slower general-purpose language model.

• Models
• Data engineering

A stateless PyTorch optimizer that normalizes gradients by their per-slice dynamic range instead of maintaining running moment statistics like Adam or RMSprop.

• Models

A 9-billion-parameter vision-language model, released with its evaluation code, that scans compressed page images and selectively expands only the relevant ones to their uncompressed form using learned tools, for answering questions over long documents.

• Models
• Papers

Downstream tasks and fine-tuning code built on the LucaOne biological foundation model's embeddings, covering protein and nucleic-acid sequence- and token-level classification and regression.

• Models

A local-first workbench for Apple Silicon that brings agent sessions, model comparison, fine-tuning, benchmarking and runtime monitoring for local MLX and remote language models into one interface.

• Models
• Inference