FlexGen
| FlexGen | minimal-llama | |
|---|---|---|
| 39 | 4 | |
| 9,007 | 456 | |
| 0.8% | - | |
| 3.0 | 8.5 | |
| 15 days ago | 7 months ago | |
| Python | Python | |
| 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.
FlexGen
- Run 70B LLM Inference on a Single 4GB GPU with This New Technique
- Colorful Custom RTX 4060 Ti GPU Clocks Outed, 8 GB VRAM Confirmed
-
Local Alternatives of ChatGPT and Midjourney
LLaMA, Pythia, RWKV, Flan-T5 (self-hosted), FlexGen
- FlexGen: Running large language models on a single GPU
-
Show HN: Finetune LLaMA-7B on commodity GPUs using your own text
> With no real knowledge of LLM and only recently started to understand what LLM terms mean, such as 'model, inference, LLM model, intruction set, fine tuning' whatelse do you think is required to make a took like yours?
This was mee a few weeks ago. I got interested in all this when FlexGen (https://github.com/FMInference/FlexGen) was announced, which allowed to run inference using OPT model on consumer hardware. I'm an avid user of Stable Diffusion, and I wanted to see if I can have an SD equivalent of ChatGPT.
Not understanding the details of hyperparameters or terminology, I basically asked ChatGPT to explain to me what these things are:
Explain to someone who is a software engineer with limited knowledge of ML terms or linear algebra, what is "feed forward" and "self-attention" in the context of ML and large language models. Provide examples when possible. - Could this new flexgen be used in place of GPTq? or is this different?
- OpenAI is expensive
minimal-llama
- Show HN: Finetune LLaMA-7B on commodity GPUs using your own text
-
Visual ChatGPT
I can't edit my comment now, but it's 30B that needs 18GB of VRAM.
LLaMA-13B, GPT-3 175B level, only needs 10GB of VRAM with the GPTQ 4bit quantization.
>do you think there's anything left to trim? like weight pruning, or LoRA, or I dunno, some kind of Huffman coding scheme that lets you mix 4-bit, 2-bit and 1-bit quantizations?
Absolutely. The GPTQ paper claims negligible output quality loss with 3-bit quantization. The GPTQ-for-LLaMA repo supports 3-bit quantization and inference. So this extra 25% savings is already possible.
As of right GPTQ-for-LLaMA is using a VRAM hungry attention method. Flash attention will reduce the requirements for 7B to 4GB and possibly fit 30B with a 2048 context window into 16GB, all before stacking 3-bit.
Pruning is a possibility but I'm not aware of anyone working on it yet.
LoRa has already been implemented. See https://github.com/zphang/minimal-llama#peft-fine-tuning-wit...
What are some alternatives?
llama - Inference code for Llama models
visual-chatgpt - Official repo for the paper: Visual ChatGPT: Talking, Drawing and Editing with Visual Foundation Models [Moved to: https://github.com/microsoft/TaskMatrix]
text-generation-webui - A Gradio web UI for Large Language Models. Supports transformers, GPTQ, AWQ, EXL2, llama.cpp (GGUF), Llama models.
whisper.cpp - Port of OpenAI's Whisper model in C/C++
text-generation-inference - Large Language Model Text Generation Inference
simple-llm-finetuner - Simple UI for LLM Model Finetuning
alpaca-lora - Instruct-tune LLaMA on consumer hardware
DeepSpeed - DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
GPTQ-for-LLaMa - 4 bits quantization of LLaMA using GPTQ
audiolm-pytorch - Implementation of AudioLM, a SOTA Language Modeling Approach to Audio Generation out of Google Research, in Pytorch