Microsoft Orca Model Download !!EXCLUSIVE!!
Tamela Mckane
Orca 2 is built for research purposes only and provides a single turn response in tasks such as reasoning over user given data, reading comprehension, math problem solving and text summarization. The model is designed to excel particularly in reasoning.
Data Biases: Large language models, trained on extensive data, can inadvertently carrybiases present in the source data. Consequently, the models may generate outputs that couldbe potentially biased or unfair.
Lack of Contextual Understanding: Despite their impressive capabilities in language understanding and generation, these models exhibit limited real-world understanding, resultingin potential inaccuracies or nonsensical responses.
Content Harms: There are various types of content harms that large language modelscan cause. It is important to be aware of them when using these models, and to takeactions to prevent them. It is recommended to leverage various content moderation servicesprovided by different companies and institutions. On an important note, we hope for betterregulations and standards from government and technology leaders around content harmsfor AI technologies in future. We value and acknowledge the important role that researchand open source community can play in this direction.
Hallucination: It is important to be aware and cautious not to entirely rely on a givenlanguage model for critical decisions or information that might have deep impact as it isnot obvious how to prevent these models from fabricating content. Moreover, it is not clearwhether small models may be more susceptible to hallucination in ungrounded generationuse cases due to their smaller sizes and hence reduced memorization capacities. This is anactive research topic and we hope there will be more rigorous measurement, understandingand mitigations around this topic.
System messages: Orca 2 demonstrates variance in performance depending on the systeminstructions. Additionally, the stochasticity introduced by the model size may lead togeneration of non-deterministic responses to different system instructions.
Zero-Shot Settings: Orca 2 was trained on data that mostly simulate zero-shot settings.While the model demonstrate very strong performance in zero-shot settings, it does not showthe same gains of using few-shot learning compared to other, specially larger, models.
Synthetic data: As Orca 2 is trained on synthetic data, it could inherit both the advantagesand shortcomings of the models and methods used for data generation. We posit that Orca2 benefits from the safety measures incorporated during training and safety guardrails (e.g.,content filter) within the Azure OpenAI API. However, detailed studies are required forbetter quantification of such risks.
This model is solely designed for research settings, and its testing has only been carriedout in such environments. It should not be used in downstream applications, as additionalanalysis is needed to assess potential harm or bias in the proposed application.
The usage of Azure AI Content Safety on top of model prediction is strongly encouraged and can help prevent content harms. Azure AI Content Safety is a content moderation platformthat uses AI to keep your content safe. By integrating Orca 2 with Azure AI Content Safety, we can moderate the model output by scanning it for sexual content, violence, hate, and self-harm with multiple severity levels and multi-lingual detection.
Microsoft Research released its Orca 2 LLM, a fine-tuned version of Llama 2 that performs as well as or better than models that contain 10x the number of parameters. Orca 2 uses a synthetic training dataset and a new technique called Prompt Erasure to achieve this performance.
Orca 2 models are trained using a teacher-student scheme, where a larger, more powerful LLM acts as a teacher for a smaller student LLM, with the goal of improving the performance of the student to be comparable with that of a larger model. Microsoft's training technique teaches the smaller model multiple reasoning techniques and also how to choose the most effective technique for a given task. To do this, the teacher is given sophisticated prompts to trigger a certain reasoning behavior. However, in a scheme called Prompt Erasure, the student is given only the task requirements and desired response, but not the teacher's prompt. When evaluated on benchmarks, a 13B parameter Orca 2 model outperformed a baseline 13B parameter Llama 2 by 47.54%. The 7B parameter Orca 2 was "better or comparable" to a 70B parameter Llama 2 on reasoning tasks.
Although LLMs like ChatGPT can often perform well on a wide range of tasks with few-shot prompting, hosting the models is challenging due to their memory and compute requirements. Smaller models can also perform well when fine-tuned, and many researchers have investigated training them with synthetic datasets generated by larger LLMs. InfoQ recently covered Google's Distilling Step-by-Step method which prompts a teacher LLM to automatically generate a small fine-tuning dataset that contains both an input with an output label, as well as a "rationale" for why the output label was chosen. InfoQ also covered Stability AI's Stable Beluga model which is trained using Microsoft's original Orca 1 scheme, which uses Explanation Tuning, where the teacher LLM is prompted to "generate detailed answers."
To evaluate the methodology, Microsoft compared Orca 2 model performance to several baseline models, including Llama 2, ChatGPT (GPT-3.5) and GPT-4. The benchmark tasks included reasoning, language understanding, text completion, and summarization. On the reasoning benchmarks, the 13B parameter Orca 2 model outperformed all baselines except ChatGPT and GPT-4. They also found that giving Orca 2 a "cautious" system prompt ("You are a cautious assistant. You carefully follow instructions.") gave it a small performance boost compared to an empty system prompt.
Many brilliant ideas are so simple...Like "Prompt Erasure" in Orca 2: Instead of presenting the entire prompt, only the task and the answer are shown to the model (it filters the full prompt used to generate those answers). It helps the model to strategize at a higher level. Such a nice paper. I highly recommend reading it all the way through.
Microsoft Research has announced the release of Orca 2, the latest iteration of their small language model series aimed at expanding the capabilities of smaller AI models. Coming in at 7 billion and 13 billion parameters, Orca 2 demonstrates advanced reasoning abilities on par with or exceeding much larger models, even those 5-10 times its size.
Orca 2 builds on the original 13B Orca model released earlier this year, which showed improved reasoning by imitating the step-by-step explanations of more powerful AI systems like GPT-4. The key to Orca 2's success lies in its training: it is fine-tuned on high-quality, synthetic data derived from the LLAMA 2 base models, a method that has proven effective in elevating its reasoning capabilities.
The research team's approach is both innovative and strategic. They have recognized that different tasks benefit from tailored solution strategies, and smaller models may need different approaches than their larger counterparts.
As such, Orca 2 has been trained on a vast dataset demonstrating various techniques like step-by-step reasoning, extraction-and-generation, and direct answering. The data was obtained from a more capable "teacher" model, which allowed Orca 2 to learn when to apply different strategies based on the problem at hand. This flexible approach is what enables Orca 2 to match or surpass much bigger models.
Comprehensive benchmarks show Orca 2 significantly outperforming other models of equivalent size on metrics related to language understanding, common sense reasoning, multi-step math problems, reading comprehension, summarization, and more. For instance, on zero-shot reasoning tasks, Orca 2-13B achieves over 25% higher accuracy than comparable 13B models and is on par with a 70B model.
While Orca 2 exhibits constraints inherited from its base LLaMA 2 model and shares limitations common to LLMs, its strong zero-shot reasoning highlights the potential for advancing smaller neural networks. Microsoft believes specialized training approaches like that used for Orca 2 can unlock new use cases balancing efficiency and capability for deployment.
Orca 2 has not undergone safety-focused RLHF tuning. However, Microsoft suggests tailored synthetic data could similarly teach safety and mitigation behaviors. They have open-sourced Orca 2 to spur further research into developing and aligning smaller but capable language models.
Despite the notable advancements made by artificial intelligence in the last decade, which include defeating human champions in strategic games like Chess and GO and predicting the 3D structure of proteins, the widespread adoption of large language models (LLMs) signifies a paradigm shift. These models, poised to transform human-computer interactions, have become indispensable across various sectors, including education, customer services, information retrieval, software development, media, and healthcare. While these technological strides unlock scientific breakthroughs and fuel industrial growth, a notable downside for the planet exists.
The process of training and utilizing LLMs consumes an immense amount of energy, resulting in a substantial environmental impact marked by an increased carbon footprint and greenhouse gas emissions. A recent study from the College of Information and Computer Sciences at the University of Massachusetts Amherst revealed that training LLMs can emit over 626,000 pounds of carbon dioxide, roughly equivalent to the lifetime emissions of five cars. Hugging Face, an AI startup, found that the training of BLOOM, a large language model launched earlier in the year, led to 25 metric tons of carbon dioxide emissions. Similarly, Facebook's AI model, Meena, accumulates a carbon footprint on par with the environmental impact of driving a car for more than 240,000 miles throughout its training process.
df19127ead