hyperlearn
unsloth
Our great sponsors
hyperlearn | unsloth | |
---|---|---|
4 | 15 | |
1,510 | 7,263 | |
0.0% | 49.5% | |
0.0 | 9.4 | |
over 1 year ago | 3 days ago | |
Jupyter Notebook | Python | |
Apache License 2.0 | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
hyperlearn
-
80% faster, 50% less memory, 0% accuracy loss Llama finetuning
I agree fully - what do you suggest then? OSS the entire code base and using AGPL3? I tried that with https://github.com/danielhanchen/hyperlearn to no avail - we couldn't even monetize it at all, so I just OSSed everything.
I listed all the research articles and methods in Hyperlearn which in the end were gobbled up by other packages.
We still have to cover life expenses and stuff sadly as a startup.
Do you have any suggestions how we could go about this? We thought maybe an actual training / inference platform, and not even OSSing any code, but we decided against this, so we OSSed some code.
Ay suggestions are welcome!
-
80% faster, 50% less memory, 0% loss of accuracy Llama finetuning
Good point - the main issue is we encountered this exact issue with our old package Hyperlearn (https://github.com/danielhanchen/hyperlearn).
I OSSed all the code to the community - I'm actually an extremely open person and I love contributing to the OSS community.
The issue was the package got gobbled up by other startups and big tech companies with no credit - I didn't want any cash from it, but it stung and hurt really bad hearing other startups and companies claim it was them who made it faster, whilst it was actually my work. It hurt really bad - as an OSS person, I don't want money, but just some recognition for the work.
I also used to accept and help everyone with their writing their startup's software, but I never got paid or even any thanks - sadly I didn't expect the world to be such a hostile place.
So after a sad awakening, I decided with my brother instead of OSSing everything, we would first OSS something which is still very good - 5X faster training is already very reasonable.
I'm all open to other suggestions on how we should approach this though! There are no evil intentions - in fact I insisted we OSS EVERYTHING even the 30x faster algos, but after a level headed discussion with my brother - we still have to pay life expenses no?
If you have other ways we can go about this - I'm all ears!! We're literally making stuff up as we go along!
-
[Project] BFLOAT16 on ALL hardware (>= 2009), up to 2000x faster ML algos, 50% less RAM usage for all old/new hardware - Hyperlearn Reborn.
Hello everyone!! It's been a while!! Years back I released Hyperlearn https://github.com/danielhanchen/hyperlearn. It has 1.2K Github stars, where I made tonnes of algos faster:
unsloth
-
Ask HN: Most efficient way to fine-tune an LLM in 2024?
Gemma 7b is 2.4x faster than HF + FA2.
Check out https://github.com/unslothai/unsloth for full benchmarks!
-
Gemma doesn't suck anymore – 8 bug fixes
Here are the missing links:
* Gemma, a family of open models from Google: https://ai.google.dev/gemma
* Unsloth is a tool/method for training models faster (IIUC): https://github.com/unslothai/unsloth
-
AMD ROCm Software Blogs
Thanks! Again, partnerships over customers. If you're experienced and have the technical chops to make a MI300x sing, we want to work with you. Our model is that we are the capex/opex investor for businesses. As much as I love software, Hot Aisle is more of a hardware business. Running super high end large scale compute is an extreme challenge in itself. We are less interested in building the software side of things and want to foster those who can focus on that side.
https://github.com/unslothai/unsloth/issues/160
https://github.com/search?q=repo%3Apredibase%2Florax+rocm&ty...
https://github.com/sgl-project/sglang/issues/157
https://github.com/casper-hansen/AutoAWQ (supports rocm)
-
Show HN: We got fine-tuning Mistral-7B to not suck
Unsloth’s colab notebooks for fine-tuning Mistral-7B are super easy to use and run fine in just about any colab instance:
https://github.com/unslothai/unsloth
It’s my default now for experimenting and basic training. If I want to get into the weeds with the training, I use axolotl, but 9/10, it’s not really necessary.
-
Mistral 7B Fine-Tune Optimized
If anyone wants to finetune their own Mistral 7b model 2.2x faster and use 62% less memory - give our open source package Unsloth a try! https://github.com/unslothai/unsloth a try! :)
-
Has anyone tried out the ASPEN-Framework for LoRA Fine-Tuning yet and can share their experience?
https://github.com/unslothai/unsloth seems good and more relevant to your aims perhaps but I haven't tried it.
-
Can we discuss MLOps, Deployment, Optimizations, and Speed?
The unsloth project offers some low-level optimizations for Llama et al, and as of today some prelim Mistral work (which I heard is the llama architecture?)
- Show HN: 80% faster, 50% less memory, 0% loss of accuracy Llama finetuning
-
80% faster, 50% less memory, 0% accuracy loss Llama finetuning
This seems to just be a link to the Unsloth Github repo[0], which in turn is the free version of Unsloth Pro/Max[1]. Maybe the link should be changed?
[0]: https://github.com/unslothai/unsloth
- 80% faster, 50% less memory, 0% loss of accuracy Llama finetuning
What are some alternatives?
gpt-fast - Simple and efficient pytorch-native transformer text generation in <1000 LOC of python.
DeepSpeed - DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
data-science-notes - Notes of IBM Data Science Professional Certificate Courses on Coursera
llama.cpp - LLM inference in C/C++
notebooks - Implement, demonstrate, reproduce and extend the results of the Risk articles 'Differential Machine Learning' (2020) and 'PCA with a Difference' (2021) by Huge and Savine, and cover implementation details left out from the papers.
nanoChatGPT - nanogpt turned into a chat model
ocaml-torch - OCaml bindings for PyTorch
DiffSharp - DiffSharp: Differentiable Functional Programming
transformers - 🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
MegEngine - MegEngine 是一个快速、可拓展、易于使用且支持自动求导的深度学习框架
uniteai - Your AI Stack in Your Editor