Methods › General › Distributed Methods › Chimera
Chimera
Introduced by Shigang Li et al. in Chimera: Efficiently Training Large-Scale Neural Networks with Bidirectional Pipelines
archive 2025-07-28 Description, source and code snippet are the archive's method entry.
Chimera is a pipeline model parallelism scheme which combines bidirectional pipelines for efficiently training large-scale models. The key idea of Chimera is to combine two pipelines in different directions (down and up pipelines).
Denote N as the number of micro-batches executed by each worker within a training iteration, and D the number of pipeline stages (depth), and P the number of workers.
The Figure shows an example with four pipeline stages (i.e. D=4). Here we assume there are D micro-batches executed by each worker within a training iteration, namely N=D, which is the minimum to keep all the stages active.
In the down pipeline, stage₀∼stage₃ are mapped to P₀∼P₃ linearly, while in the up pipeline the stages are mapped in a completely opposite order. The N (assuming an even number) micro-batches are equally partitioned among the two pipelines. Each pipeline schedules N/2 micro-batches using 1F1B strategy, as shown in the left part of the Figure. Then, by merging these two pipelines together, we obtain the pipeline schedule of Chimera. Given an even number of stages D (which can be easily satisfied in practice), it is guaranteed that there is no conflict (i.e., there is at most one micro-batch occupies the same time slot on each worker) during merging.
Papers archive 2025-07-28
16 shown of 16, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.
-
CHIMERA: A Knowledge Base of Idea Recombination in Scientific Literature 27 May 2025 · 1 repository · arXiv:2505.20779
-
A CMOS Probabilistic Computing Chip With In-situ hardware Aware Learning 18 Apr 2025 · 0 repositories · arXiv:2504.14070
-
Chimera: A Block-Based Neural Architecture Search Framework for Event-Based Object Detection 27 Dec 2024 · 0 repositories · arXiv:2412.19646
-
Language model driven: a PROTAC generation pipeline with dual constraints of structure and property 12 Dec 2024 · 0 repositories · arXiv:2412.09661
-
Chimera: Improving Generalist Model with Domain-Specific Experts 8 Dec 2024 · 0 repositories · arXiv:2412.05983
-
Chimera: Accurate retrosynthesis prediction by ensembling models with diverse inductive biases 6 Dec 2024 · 0 repositories · arXiv:2412.05269
-
Chimera: Effectively Modeling Multivariate Time Series with 2-Dimensional State Space Models 6 Jun 2024 · 0 repositories · arXiv:2406.04320
-
Chimera: A Lossless Decoding Method for Accelerating Large Language Models Inference by Fusing all Tokens 24 Feb 2024 · 0 repositories · arXiv:2402.15758
-
Artificial Bee Colony optimization of Deep Convolutional Neural Networks in the context of Biomedical Imaging 23 Feb 2024 · 0 repositories · arXiv:2402.15246
-
Spoofing-Resilient LiDAR-GPS Factor Graph Localization with Chimera Authentication 10 Jul 2023 · 1 repository · arXiv:2307.04692
-
Investigating the generative dynamics of energy-based neural networks 11 May 2023 · 0 repositories · arXiv:2305.06745
-
Chimera: A Hybrid Machine Learning Driven Multi-Objective Design Space Exploration Tool for FPGA High-Level Synthesis 3 Jul 2022 · 0 repositories · arXiv:2207.07917
-
Physics-inspired Ising Computing with Ring Oscillator Activated p-bits 15 May 2022 · 0 repositories · arXiv:2205.07402
-
Complex dynamics of a heterogeneous network of Hindmarsh-Rose neurons 3 May 2022 · 0 repositories · arXiv:2205.01790
-
Convex Non-negative Matrix Factorization Through Quantum Annealing 28 Mar 2022 · 0 repositories · arXiv:2203.15634
-
Chimera: Efficiently Training Large-Scale Neural Networks with Bidirectional Pipelines 14 Jul 2021 · 1 repository · arXiv:2107.06925
Tasks archive 2025-07-28
20 shown of 29 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.
Usage over time archive 2025-07-28
Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).
Categories archive 2025-07-28
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections