plutonium-239/memsave_torch
Lowering PyTorch's Memory Consumption for Selective Differentiation
This package helps deep learning engineers and researchers manage GPU memory more efficiently when training large PyTorch models. It takes existing neural network layers and converts them into memory-saving versions. The output is a functionally identical model that consumes less memory, especially useful when only updating a subset of the model's parameters.
No commits in the last 6 months.
Use this if you are a deep learning practitioner training PyTorch models and frequently encounter 'out of memory' errors, particularly when fine-tuning or using techniques where only certain layers or parameters require gradient computation.
Not ideal if your PyTorch models are small, you always train all parameters, or you are not experiencing memory consumption issues on your current hardware.
Stars
12
Forks
1
Language
Python
License
MIT
Category
Last pushed
Aug 29, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/plutonium-239/memsave_torch"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
unslothai/unsloth
Fine-tuning & Reinforcement Learning for LLMs. 🦥 Train OpenAI gpt-oss, DeepSeek, Qwen, Llama,...
huggingface/peft
🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.
modelscope/ms-swift
Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.5, DeepSeek-R1, GLM-5,...
oumi-ai/oumi
Easily fine-tune, evaluate and deploy gpt-oss, Qwen3, DeepSeek-R1, or any open source LLM / VLM!
linkedin/Liger-Kernel
Efficient Triton Kernels for LLM Training