The BigCode community, an open-scientific collaboration working on the responsible development of Large Language Models for Code (Code LLMs), introduces StarCoder and StarCoderBase: 15.5B parameter models with 8K context length, infilling capabilities and fast large-batch inference enabled by multi-query attention. StarCoderBase is trained on 1 trillion tokens sourced from The Stack, a large collection of permissively licensed GitHub repositories with inspection tools and an opt-out process. We fine-tuned StarCoderBase on 35B Python tokens, resulting in the creation of StarCoder. We perform the most comprehensive evaluation of Code LLMs to date and show that StarCoderBase outperforms every open Code LLM that supports multiple programming languages and matches or outperforms the OpenAI code-cushman-001 model. Furthermore, StarCoder outperforms every model that is fine-tuned on Python, can be prompted to achieve 40\% pass@1 on HumanEval, and still retains its performance on other programming languages. We take several important steps towards a safe open-access model release, including an improved PII redaction pipeline and a novel attribution tracing tool, and make the StarCoder models publicly available under a more commercially viable version of the Open Responsible AI Model license.
Learning the hash representation of multi-view heterogeneous data is an important task in multimedia retrieval. However, existing methods fail to effectively fuse the multi-view features and utilize the metric information provided by the dissimilar samples, leading to limited retrieval precision. Current methods utilize weighted sum or concatenation to fuse the multi-view features. We argue that these fusion methods cannot capture the interaction among different views. Furthermore, these methods ignored the information provided by the dissimilar samples. We propose a novel deep metric multi-view hashing (DMMVH) method to address the mentioned problems. Extensive empirical evidence is presented to show that gate-based fusion is better than typical methods. We introduce deep metric learning to the multi-view hashing problems, which can utilize metric information of dissimilar samples. On the MIR-Flickr25K, MS COCO, and NUS-WIDE, our method outperforms the current state-of-the-art methods by a large margin (up to 15.28 mean Average Precision (mAP) improvement).
AI-driven medical history-taking is an important component in symptom checking, automated patient intake, triage, and other AI virtual care applications. As history-taking is extremely varied, machine learning models require a significant amount of data to train. To overcome this challenge, existing systems are developed using indirect data or expert knowledge. This leads to a training-inference gap as models are trained on different kinds of data than what they observe at inference time. In this work, we present a two-stage re-ranking approach that helps close the training-inference gap by re-ranking the first-stage question candidates using a dialogue-contextualized model. For this, we propose a new model, global re-ranker, which cross-encodes the dialogue with all questions simultaneously, and compare it with several existing neural baselines. We test both transformer and S4-based language model backbones. We find that relative to the expert system, the best performance is achieved by our proposed global re-ranker with a transformer backbone, resulting in a 30% higher normalized discount cumulative gain (nDCG) and a 77% higher mean average precision (mAP).
As language models grow ever larger, the need for large-scale high-quality text datasets has never been more pressing, especially in multilingual settings. The BigScience workshop, a 1-year international and multidisciplinary initiative, was formed with the goal of researching and training large language models as a values-driven undertaking, putting issues of ethics, harm, and governance in the foreground. This paper documents the data creation and curation efforts undertaken by BigScience to assemble the Responsible Open-science Open-collaboration Text Sources (ROOTS) corpus, a 1.6TB dataset spanning 59 languages that was used to train the 176-billion-parameter BigScience Large Open-science Open-access Multilingual (BLOOM) language model. We further release a large initial subset of the corpus and analyses thereof, and hope to empower large-scale monolingual and multilingual modeling projects with both the data and the processing tools, as well as stimulate research around this large multilingual corpus.
A framework performing Visual Commonsense Reasoning(VCR) needs to choose an answer and further provide a rationale justifying based on the given image and question, where the image contains all the facts for reasoning and requires to be sufficiently understood. Previous methods use a detector applied on the image to obtain a set of visual objects without considering the exact positions of them in the scene, which is inadequate for properly understanding spatial and semantic relationships between objects. In addition, VCR samples are quite diverse, and parameters of the framework tend to be trained suboptimally based on mini-batches. To address above challenges, pseudo 3D perception Transformer with multi-level confidence optimization named PPTMCO is proposed for VCR in this paper. Specifically, image depth is introduced to represent pseudo 3-dimension(3D) positions of objects along with 2-dimension(2D) coordinates in the image and further enhance visual features. Then, considering that relationships between objects are influenced by depth, depth-aware Transformer is proposed to do attention mechanism guided by depth differences from answer words and objects to objects, where each word is tagged with pseudo depth value according to related objects. To better optimize parameters of the framework, a model parameter estimation method is further proposed to weightedly integrate parameters optimized by mini-batches based on multi-level reasoning confidence. Experiments on the benchmark VCR dataset demonstrate the proposed framework performs better against the state-of-the-art approaches.
Large language models (LLMs) have been shown to be able to perform new tasks based on a few demonstrations or natural language instructions. While these capabilities have led to widespread adoption, most LLMs are developed by resource-rich organizations and are frequently kept from the public. As a step towards democratizing this powerful technology, we present BLOOM, a 176B-parameter open-access language model designed and built thanks to a collaboration of hundreds of researchers. BLOOM is a decoder-only Transformer language model that was trained on the ROOTS corpus, a dataset comprising hundreds of sources in 46 natural and 13 programming languages (59 in total). We find that BLOOM achieves competitive performance on a wide variety of benchmarks, with stronger results after undergoing multitask prompted finetuning. To facilitate future research and applications using LLMs, we publicly release our models and code under the Responsible AI License.
Inducing semantic representations directly from speech signals is a highly challenging task but has many useful applications in speech mining and spoken language understanding. This study tackles the unsupervised learning of semantic representations for spoken utterances. Through converting speech signals into hidden units generated from acoustic unit discovery, we propose WavEmbed, a multimodal sequential autoencoder that predicts hidden units from a dense representation of speech. Secondly, we also propose S-HuBERT to induce meaning through knowledge distillation, in which a sentence embedding model is first trained on hidden units and passes its knowledge to a speech encoder through contrastive learning. The best performing model achieves a moderate correlation (0.5~0.6) with human judgments, without relying on any labels or transcriptions. Furthermore, these models can also be easily extended to leverage textual transcriptions of speech to learn much better speech embeddings that are strongly correlated with human annotations. Our proposed methods are applicable to the development of purely data-driven systems for speech mining, indexing and search.
In this study, we tackle massively multilingual grapheme-to-phoneme conversion through implementing G2P models based on ByT5. We have curated a G2P dataset from various sources that covers around 100 languages and trained large-scale multilingual G2P models based on ByT5. We found that ByT5 operating on byte-level inputs significantly outperformed the token-based mT5 model in terms of multilingual G2P. Pairwise comparison with monolingual models in these languages suggests that multilingual ByT5 models generally lower the phone error rate by jointly learning from a variety of languages. The pretrained model can further benefit low resource G2P through zero-shot prediction on unseen languages or provides pretrained weights for finetuning, which helps the model converge to a lower phone error rate than randomly initialized weights. To facilitate future research on multilingual G2P, we make available our code and pretrained multilingual G2P models at: https://github.com/lingjzhu/CharsiuG2P.
Click-through rate (CTR) Prediction is of great importance in real-world online ads systems. One challenge for the CTR prediction task is to capture the real interest of users from their clicked items, which is inherently biased by presented positions of items, i.e., more front positions tend to obtain higher CTR values. A popular line of existing works focuses on explicitly estimating position bias by result randomization which is expensive and inefficient, or by inverse propensity weighting (IPW) which relies heavily on the quality of the propensity estimation. Another common solution is modeling position as features during offline training and simply adopting fixed value or dropout tricks when serving. However, training-inference inconsistency can lead to sub-optimal performance. Furthermore, post-click information such as position values is informative while less exploited in CTR prediction. This work proposes a simple yet efficient knowledge distillation framework to alleviate the impact of position bias and leverage position information to improve CTR prediction. We demonstrate the performance of our proposed method on a real-world production dataset and online A/B tests, achieving significant improvements over competing baseline models. The proposed method has been deployed in the real world online ads systems, serving main traffic on one of the world's largest e-commercial platforms.
Learning to capture feature relations effectively and efficiently is essential in click-through rate (CTR) prediction of modern recommendation systems. Most existing CTR prediction methods model such relations either through tedious manually-designed low-order interactions or through inflexible and inefficient high-order interactions, which both require extra DNN modules for implicit interaction modeling. In this paper, we proposed a novel plug-in operation, Dynamic Parameterized Operation (DPO), to learn both explicit and implicit interaction instance-wisely. We showed that the introduction of DPO into DNN modules and Attention modules can respectively benefit two main tasks in CTR prediction, enhancing the adaptiveness of feature-based modeling and improving user behavior modeling with the instance-wise locality. Our Dynamic Parameterized Networks significantly outperforms state-of-the-art methods in the offline experiments on the public dataset and real-world production dataset, together with an online A/B test. Furthermore, the proposed Dynamic Parameterized Networks has been deployed in the ranking system of one of the world's largest e-commerce companies, serving the main traffic of hundreds of millions of active users.