Training data is a pivotal resource for building large language models (LLMs), but unfair pricing in data markets poses a serious challenge for both data buyers (e.g., LLM builders) and sellers (e.g., human annotators), which discourages market participation, reducing data quantity and quality. In this paper, we propose a fairshare pricing framework that sets training data prices using data valuation methods to quantify their contribution to LLMs. In our framework, buyers make purchasing decisions using data valuation and sellers set prices to maximize their profits based on the anticipated buyer purchases. We theoretically show that pricing derived from our framework is tightly linked to data valuation and buyers' budget, optimal for both buyers and sellers. Through market simulations using current LLMs and datasets (math problems, medical diagnosis, and physical reasoning), we show that our framework is fairshare for buyers by ensuring their purchased data is reflective of model training value, leading to higher LLM task performances per-dollar spent on data, and fairshare for sellers by ensuring they sell their data at optimal prices. Our framework lays the foundation for future research on equitable and sustainable data markets for large-scale AI.

Data valuation quantifies the value of training data, and is used for data attribution (i.e., determining the contribution of training data towards model predictions), and data selection; both of which are important for curating high-quality datasets to train large language models. In our paper, we show that data valuation through in-context probing (i.e., prompting a LLM) approximates influence functions for selecting training data. We provide a theoretical sketch on this connection based on transformer models performing "implicit" gradient descent on its in-context inputs. Our empirical findings show that in-context probing and gradient-based influence frameworks are similar in how they rank training data. Furthermore, fine-tuning experiments on data selected by either method reveal similar model performance.

In recent years, language models such as GPT-3 [5] have grown larger, and their performance on downstream natural language processing (NLP) tasks has significantly improved in low-resource settings where only a few instances per task are available (few-shot). The larger these models are, the higher their performances trend on tasks such as language generation and evaluation [39]. They can generate coherent, fluent and interesting responses. However, they can also produce responses that are repetitive and un-engaging [29], in addition to being hard to control. Dialog evaluation is the task of assessing the quality of responses generated by dialog models in terms of properties like those mentioned above. However, one significant impediment for open-domain dialog generation research is the lack of meaningful automatic metrics for open-domain dialog evaluation. Standard language generation metrics have been shown to be ineffective for dialog evaluation [11], a large part of which is because conversations can be followed by multiple valid responses.

Static datasets are ineffective for both evaluation and optimization. The Alexa Prize challenge (Ram et al., 2018; This has led to the creation of the DialPort Khatri et al., 2018) allows university teams to build Portal, which facilitates the collection of socialbots that are assessed in interactive settings flexible and evolving data as well as interactive assessment with Alexa users.