Competition
NeurIPS Large Language Model Efficiency Challenge: 1 LLM + 1GPU + 1Day
Mark Saroufim · Weiwei Yang · Christian Puhrsch · Luca Antiga · Greg Bowyer · Driss Guessous · Artidoro Pagnoni · Supriya Rao · Joseph Isaacson · Vicki Boykis · Geeta Chauhan · aaron gonzales · Davide Eynard
Room 356
Large Language Models (LLMs) have been pivotal in the recent Cambrian explosion of generative AI applications. However, existing efforts to democratize access to fine-tune and query LLMs have been largely limited by growing hardware costs required to adapt and serve these models. Enabling low cost and efficient LLM fine-tuning and inference can have significant impact on industrial and scientific applications. Here, we present a single GPU fine-tuning and inference competition. Our goal is to accelerate the development of practical software methods to reduce the costs associated with utilizing LLMs. Furthermore, by advocating for goal-oriented and infrastructure-focused evaluation frameworks that stress reproducibility, our aim is to democratize access to these methods and enhance their accessibility to the wider public.
Schedule
Fri 11:30 a.m. - 11:45 a.m.
|
Kick-Off to Efficiency: Welcoming statement for the organizers
(
Speak
)
>
SlidesLive Video Welcome statement, introduce the competition -- NeurIPS Large Language Model Efficiency Challenge: 1 LLM + 1GPU + 1Day |
Mark Saroufim · Weiwei Yang 🔗 |
Fri 11:45 a.m. - 12:00 p.m.
|
Invited Speaker: Jeremy Howard-Lessons from 25 years of machine learning competitions
(
In-person presentation
)
>
SlidesLive Video Before Kaggle, before the $1m Netflix prize, there was the KDD Cup, the first big annual machine learning competition. Starting in 1997, it brought together each year the world’s best predictive modelers. We now have over 25 years of experience of ML competitions to draw on, with tens of millions of dollars in prizes awarded. I will summarize what we’ve learned from this experience, and explain where, how, and why ML competitions can help advance machine learning research and practice. |
🔗 |
Fri 12:00 p.m. - 12:15 p.m.
|
Invited Speaker: Sebastian Raschka (lightning.ai) - LoRA in Action: Insights from Finetuning LLMs with Low-Rank Adaptation
(
In-person presentation
)
>
SlidesLive Video Low-rank adaptation (LoRA) stands as one of the most popular and effective methods for efficiently training custom Large Language Models (LLMs). As practitioners of open-source LLMs, we regard LoRA as a crucial technique in our toolkit. In this talk, I will delve into some practical insights gained from running hundreds of experiments with LoRA, addressing questions such as: How much can I save with quantized LoRA? Are Adam optimizers memory-intensive? Should we train for multiple epochs? How do we choose the LoRA rank? Moreover, the talk will include ideas for future experiments and talking points to stimulate discussions in the workshop, such as mechanisms to avoid overfitting in LoRA and strategies for combining LoRA weights from multiple experiments. |
Sebastian Raschka 🔗 |
Fri 12:15 p.m. - 12:30 p.m.
|
Unveiling Success: A100 track Team percent_bdf's Winning Strategies
(
Zoom presentation
)
>
SlidesLive Video In this session, A100 winning team Percent_bdf will reveal the strategies and work that led them to triumph, offering insights and lessons from their experience. |
Ao Liu 🔗 |
Fri 12:30 p.m. - 12:45 p.m.
|
Invited Speaker: Tim Dettmers QLoRA
(
In-person presentation
)
>
SlidesLive Video |
Tim Dettmers 🔗 |
Fri 12:45 p.m. - 1:00 p.m.
|
Invited Speaker: Sourab Mangrulka -- Generative AI for Al: 🤗 PEFT: Finetuning made simple, efficient and extendable
(
Zoom presentation
)
>
SlidesLive Video Generative AI is now becoming part and parcel of everyone’s daily life. Large Langage Models such as ChatGPT/GPT4, PaLM, Claude, Llama, Mistral, Falcon and StarCoder are at the core of this owing to their state-of-the-art performance at various Natural Language Processing (NLP) tasks, conversational skills and logical reasoning/coding. The conventional paradigm followed is to pretrain the model on web-scale data followed by finetuning on downstream tasks to get the best performance. The finetuning step becomes infeasible as models get larger due to insufficient access to dedicated hardware thereby preventing widespread availability and usage of these models. Parameter-Efficient Fine-tuning (PEFT) methods enable efficient adaptation of pre-trained language models to various downstream applications without fine-tuning all the model's parameters while maintaining performance. 🤗 PEFT is an open-source project with the vision to democratize access to fine-tuning large AI models on consumer hardware/low-resources while being simple, efficient and adaptable at scale. Here, I will present the development and design considerations that went into building 🤗 PEFT and how it fits in the Generative AI landscape. |
SOURAB MANGRULKAR 🔗 |
Fri 1:00 p.m. - 1:15 p.m.
|
Coffee break
(
break
)
>
|
🔗 |
Fri 1:15 p.m. - 1:30 p.m.
|
Unveiling Success: 4090 Track winning team's strategies
(
In-person presentation
)
>
SlidesLive Video In this session, the winning team of 4090 track will reveal the innovative strategies and teamwork that led them to triumph, offering insights and lessons from their experience. |
🔗 |
Fri 1:30 p.m. - 1:45 p.m.
|
Invited Speaker: Keming Lu (Alibaba Research) - Qwen: Towards a Generalist Model
(
In-person presentation
)
>
SlidesLive Video We introduce the large language and multimodal model series Qwen, published and opensourced by Alibaba Group. The Qwen model have achieved competitive performance against both opensource and proprietary LLMs and LMMs in both benchmark and human evaluation. This talk provides a brief overview of the model series and delves into details about building the LLMs, including pretraining, alignment, as well as the opensource. Additionally, it points out the limitations, and discusses the future work for both research community and industry in this field. |
Keming Lu 🔗 |
Fri 1:45 p.m. - 2:00 p.m.
|
Invited Speaker: Mojan Javaheripi (Microsoft Research) - Unleashing the power of Small Language Models
(
Zoom presentation
)
>
SlidesLive Video Over the past few months, we have released a suite of small language models (SLMs) called “Phi” that achieve unprecedented performance on a variety of benchmarks. Our first model, the 1.3 billion parameter Phi-1, achieved state-of-the-art performance on Python coding among SLMs. We then extended our focus to common sense reasoning and language understanding, and created a new 1.3 billion parameter model named Phi-1.5, with performance comparable to models 5x larger. Our latest model, the 2.7 billion parameter Phi-2, surpasses Phi-1.5 performance on all benchmarks, thanks to new innovations in model scaling and training data curation. In this talk, I will introduce Phi SLMs and discuss two key insights driving their performance: 1) generation and utilization of data with "textbook quality" to elevate the learning process in contrast to conventional web data, and 2) incorporation of best practices for scaling up to enhance overall performance. |
Mojan Javaheripi 🔗 |
Fri 2:00 p.m. - 2:15 p.m.
|
Invited Speaker: Leshem Choshen (IBM Research) - Efficient Evaluation for Efficient Training
(
In-person presentation
)
>
SlidesLive Video Two competing forces are ignored in evaluation: reliability and efficiency. The talk would explain the basics of the open evaluation (HELM) and the analysis done to change an awfully slow evaluation (4K GPU hours for a single model), to hundreds times faster evaluation that you can still rely on its scores. In short, maximize the variability of the data (more datasets prompts, less examples and repetitions) and give more resources to cases you care about (e.g., remove the bottom models after a fast evaluation). The analysis and more details on how to make-check smart evaluation decisions: "efficient benchmarking (of language models)" https://arxiv.org/abs/2308.11696v3 |
Leshem Choshen 🔗 |
Fri 2:15 p.m. - 2:30 p.m.
|
Award celemony and open floor discussion
(
panel discussion
)
>
SlidesLive Video Please join us to celebrate the community's achievements and the successes of this competition. The event will feature an open floor discussion on the future of the competition. We look forward to celebrating together and exploring what lies ahead. |
Weiwei Yang · Mark Saroufim · Christian Puhrsch · Joseph Isaacson · Vicki Boykis 🔗 |