Simple and efficient DeepSeek V3 SFT using pipeline parallel and expert parallel, with both FP8 and BF16 trainings
☆117Jul 27, 2025Updated last year
Alternatives and similar repositories for pipelining-sft
Users that are interested in pipelining-sft are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- A fast, local, and secure approach for training LLMs for coding tasks using GRPO with WebAssembly and interpreter feedback.☆42Apr 4, 2025Updated last year
- ☆37Oct 4, 2026Updated last week
- A scalable asynchronous reinforcement learning implementation with in-flight weight updates.☆435Aug 5, 2026Updated 2 months ago
- PyTorch Single Controller☆1,079Updated this week
- ☆23Jun 12, 2025Updated last year
- Deploy on Railway without the complexity - Free Credits Offer • AdConnect your repo and Railway handles the rest with instant previews. Quickly provision container image services, databases, and storage volumes.
- PyTorch-native post-training at scale☆705Sep 9, 2026Updated last month
- letta integration for terminalbench (#1 open source agent, in under 200 lines of code)☆19Oct 22, 2025Updated 11 months ago
- Train your own SOTA deductive reasoning model☆111Mar 6, 2025Updated last year
- [ICML 2026 Spotlight] Official implementation of TetraJet-v2: Accurate NVFP4 Training for LLMs, with fully-NVFP4 linear layer with unbias…☆18Jul 3, 2026Updated 3 months ago
- Lego for GRPO☆30May 27, 2025Updated last year
- An agent harness that compiles a model into one provably-correct, self-retargeting CUDA megakernel and self-tunes it past cuBLAS at batch…☆151Sep 18, 2026Updated 3 weeks ago
- A pure-Python Beaker client☆17Oct 14, 2025Updated last year
- PyTorch native quantization for training and inference☆3,004Updated this week
- Fault tolerance for PyTorch (HSDP, LocalSGD, DiLoCo, Streaming DiLoCo)☆547Updated this week
- Wordpress hosting with auto-scaling - Free Trial Offer • AdFully Managed hosting for WordPress and WooCommerce businesses that need reliable, auto-scalable performance. Cloudways SafeUpdates now available.
- Agentic RL Training at Scale☆2,165Updated this week
- Accelerating MoE with IO and Tile-aware Optimizations☆780Aug 29, 2026Updated last month
- Checkpoint-engine is a simple middleware to update model weights in LLM inference engines☆1,007Sep 15, 2026Updated last month
- Scalable toolkit for efficient model reinforcement☆2,067Updated this week
- The simplest, fastest repository for training/finetuning medium-sized GPTs.☆205Jan 19, 2026Updated 8 months ago
- A subset of PyTorch's neural network modules, written in Python using OpenAI's Triton.☆605Aug 14, 2026Updated 2 months ago
- Python package for compressing floating-point PyTorch tensors☆14Jul 22, 2024Updated 2 years ago
- ☆95Jul 5, 2024Updated 2 years ago
- NanoGPT speedrun in JAX. Originally at https://nor-git.pages.dev/modded-nanogpt-jax/☆17Aug 28, 2025Updated last year
- Serverless GPU API endpoints on Runpod - Get Bonus Credits • AdSkip the infrastructure headaches. Auto-scaling, pay-as-you-go, no-ops approach lets you focus on innovating your application.
- Expert Specialization MoE Solution based on CUTLASS☆27Apr 14, 2026Updated 6 months ago
- Best practices for training DeepSeek, Mixtral, Qwen and other MoE models using Megatron Core.☆204May 29, 2026Updated 4 months ago
- ☆69May 23, 2025Updated last year
- ☆58Mar 14, 2025Updated last year
- The official repo for "OpenMoE 2: Sparse Diffusion Language Models".☆58Dec 28, 2025Updated 9 months ago
- A Triton-only attention backend for vLLM☆28Jul 14, 2026Updated 3 months ago
- Vortex: Programmable Sparse Attention for Agents as Algorithm Designers☆69Aug 22, 2026Updated last month
- FlexAttention w/ FlashAttention3 Support☆27Oct 5, 2024Updated 2 years ago
- Auto math prover.☆10Jul 10, 2024Updated 2 years ago
- 1-Click AI Models by DigitalOcean Gradient • AdDeploy popular AI models on DigitalOcean Gradient GPU virtual machines with just a single click. Zero configuration with optimized deployments.
- ☆389Jan 28, 2026Updated 8 months ago
- slime is an LLM post-training framework for RL Scaling.☆8,661Updated this week
- Manages vllm-nccl dependency☆19Jun 3, 2024Updated 2 years ago
- TreeRL: LLM Reinforcement Learning with On-Policy Tree Search in ACL'25☆106Jun 16, 2025Updated last year
- ByteCheckpoint: An Unified Checkpointing Library for LFMs☆290Feb 2, 2026Updated 8 months ago
- ☆484Nov 25, 2025Updated 10 months ago
- NSA Triton Kernels written with GPT5 and Opus 4.1☆71Aug 12, 2025Updated last year