martin-marek / batch-sizeLinks
πSmall Batch Size Training for Language Models
β57Updated last week
Alternatives and similar repositories for batch-size
Users that are interested in batch-size are comparing it to the libraries listed below
Sorting:
- The simplest, fastest repository for training/finetuning medium-sized GPTs.β156Updated 2 months ago
- supporting pytorch FSDP for optimizersβ84Updated 8 months ago
- β87Updated last year
- β85Updated last year
- β57Updated 11 months ago
- Implementations of attention with the softpick function, naive and FlashAttention-2β83Updated 4 months ago
- β34Updated last year
- Explorations into the recently proposed Taylor Series Linear Attentionβ100Updated last year
- Code for NeurIPS 2024 Spotlight: "Scaling Laws and Compute-Optimal Training Beyond Fixed Training Durations"β82Updated 10 months ago
- Pytorch implementation of the PEER block from the paper, Mixture of A Million Experts, by Xu Owen He at Deepmindβ127Updated last year
- Normalized Transformer (nGPT)β187Updated 9 months ago
- Tree Attention: Topology-aware Decoding for Long-Context Attention on GPU clustersβ129Updated 9 months ago
- β53Updated last year
- Minimal (400 LOC) implementation Maximum (multi-node, FSDP) GPT trainingβ131Updated last year
- β27Updated last year
- β32Updated 9 months ago
- A MAD laboratory to improve AI architecture designs π§ͺβ127Updated 8 months ago
- β53Updated last year
- Custom triton kernels for training Karpathy's nanoGPT.β19Updated 10 months ago
- β65Updated 9 months ago
- β11Updated 6 months ago
- Understand and test language model architectures on synthetic tasks.β222Updated last month
- β102Updated last month
- Minimal but scalable implementation of large language models in JAXβ35Updated this week
- [ICLR 2025] Official PyTorch Implementation of Gated Delta Networks: Improving Mamba2 with Delta Ruleβ209Updated 5 months ago
- Mixture of A Million Expertsβ46Updated last year
- Tiny re-implementation of MDM in style of LLaDA and nano-gpt speedrunβ56Updated 5 months ago
- β116Updated 2 months ago
- Why Do We Need Weight Decay in Modern Deep Learning? [NeurIPS 2024]β67Updated 11 months ago
- A basic pure pytorch implementation of flash attentionβ16Updated 10 months ago