Since developing a new multimodal LLM (MLLM) by pre-training on tremendous image-text pairs from scratch can be exceedingly resource-consuming, connecting an existing LLM with a comparatively lightweight visual prompt generator (VPG) becomes a feasible paradigm.

Since developing a new multimodal LLM (MLLM) by pre-training on tremendous image-text pairs from scratch can be exceedingly resource-consuming, connecting an existing LLM with a comparatively lightweight visual prompt generator (VPG) becomes a feasible paradigm. An alternative solution is transferring an existing VPG from one MLLM to the target MLLM. In this work, we investigate VPG transferability across LLMs for the first time, aiming to reduce the cost of VPG training. Specifically, we explore VPG transfer across different LLM sizes (e.g., small-to-large) and types. We identify key factors to maximize transfer efficiency, based on which we develop a simple yet highly effective two-stage transfer framework, called VPGTrans. Notably, it enables VPG transfer from BLIP-2 OPT 2.7B to BLIP-2 OPT 6.7B with less than 10% of the GPU hours using only 10.7% of the training data compared to training a VPG for OPT 6.7B from scratch.

Post-Training Quantization (PTQ) enhances the efficiency of Large Language Models (LLMs) by enabling faster operation and compatibility with more accessible hardware through reduced memory usage, at the cost of small performance drops. We explore the role of calibration sets in PTQ, specifically their effect on hidden activations in various notable open-source LLMs. Calibration sets are crucial for evaluating activation magnitudes and identifying outliers, which can distort the quantization range and negatively impact performance. Our analysis reveals a marked contrast in quantization effectiveness across models. The older OPT model, upon which much of the quantization literature is based, shows significant performance deterioration and high susceptibility to outliers with varying calibration sets. In contrast, newer models like Llama-2 7B, Llama-3 8B, Command-R 35B, and Mistral 7B demonstrate strong robustness, with Mistral 7B showing near-immunity to outliers and stable activations. These findings suggest a shift in PTQ strategies might be needed. As advancements in pre-training methods reduce the relevance of outliers, there is an emerging need to reassess the fundamentals of current quantization literature. The emphasis should pivot towards optimizing inference speed, rather than primarily focusing on outlier preservation, to align with the evolving characteristics of state-of-the-art LLMs.

The wide acceptance of large language models (LLMs) has unlocked new applications and social risks. Popular countermeasures aim at detecting misinformation, usually involve domain specific models trained to recognize the relevance of any information. Instead of evaluating the validity of the information, we propose to investigate LLM generated text from the perspective of trust. In this study, we define trust as the ability to know if an input text was generated by a LLM or a human. To do so, we design SPOT, an efficient method, that classifies the source of any, standalone, text input based on originality score. This score is derived from the prediction of a given LLM to detect other LLMs. We empirically demonstrate the robustness of the method to the architecture, training data, evaluation data, task and compression of modern LLMs.

Large language models demonstrate a remarkable capability for learning to solve new tasks from a few examples. The prompt template, or the way the input examples are formatted to obtain the prompt, is an important yet often overlooked aspect of in-context learning. In this work, we conduct a comprehensive study of the template format's influence on the in-context learning performance. We evaluate the impact of the prompt template across models (from 770M to 70B parameters) and 4 standard classification datasets. We show that a poor choice of the template can reduce the performance of the strongest models and inference methods to a random guess level. More importantly, the best templates do not transfer between different setups and even between models of the same family. Our findings show that the currently prevalent approach to evaluation, which ignores template selection, may give misleading results due to different templates in different works. As a first step towards mitigating this issue, we propose Template Ensembles that aggregate model predictions across several templates. This simple test-time augmentation boosts average performance while being robust to the choice of random set of templates.

Large language models (LLMs) are central to modern natural language processing, delivering exceptional performance in various tasks. However, their substantial computational and memory requirements present challenges, especially for devices with limited DRAM capacity. This paper tackles the challenge of efficiently running LLMs that exceed the available DRAM capacity by storing the model parameters in flash memory, but bringing them on demand to DRAM. Our method involves constructing an inference cost model that takes into account the characteristics of flash memory, guiding us to optimize in two critical areas: reducing the volume of data transferred from flash and reading data in larger, more contiguous chunks. Within this hardware-informed framework, we introduce two principal techniques. First, "windowing" strategically reduces data transfer by reusing previously activated neurons, and second, "row-column bundling", tailored to the sequential data access strengths of flash memory, increases the size of data chunks read from flash memory. These methods collectively enable running models up to twice the size of the available DRAM, with a 4-5x and 20-25x increase in inference speed compared to naive loading approaches in CPU and GPU, respectively. Our integration of sparsity awareness, context-adaptive loading, and a hardware-oriented design paves the way for effective inference of LLMs on devices with limited memory.

Low-rank adaptation (LoRA) and question-answer datasets from large language models have made it much easier for much smaller models to be finetuned to the point where they display sophisticated conversational abilities. In this paper, we present Eluwa, a family of LoRA models that use the Stanford Alpaca dataset and massively improve the capabilities of Facebook's OPT 1.3B, 2.7B and 6.7B models. We benchmark these models in multiple ways, including letting GPT-4 judge their answers to prompts that span general knowledge, writing, programming and other tasks. We show that smaller models here can be fine-tuned to be as performant as models 3x larger - all for as little as 40 USD in compute.