LyRise, a recruitment platform connecting companies with AI experts from the Middle East and North Africa, uses Meta's Llama 2 and 3 to implement AI-driven candidate matching. By utilizing Llama's text summarization and RAG capabilities to condense job descriptions and resumes, LyRise reduces time-to-hire by 50%, while ensuring quality talent acquisition.

Gupshup.io's CX Product Suite 'Converse' is now powered by ACE LLM, domain-specific models fine-tuned for enterprise-grade conversations based on Meta's Llama 2. This enhances their offerings in customer acquisition, marketing, commerce, and support.

Niantic, known for augmented reality games like Pokémon GO, uses Meta's Llama 2 to develop Peridot, a mobile AR game featuring virtual pets called "Dots." By integrating Llama 2, Dots exhibit dynamic and responsive behaviors, making interactions unique and enhancing user engagement. Additionally, Niantic employs computer vision with the Niantic Lightship ARDK to enable Dots to interact with the physical environment, creating a more immersive and lifelike gaming experience. This AI-driven approach accelerates development, reduces manual programming, and delivers personalized interactions, improving player satisfaction and engagement.

Sarvam AI specializes in developing AI solutions for Indic languages. They developed a Hindi Large Language Model by extending Meta's Llama 2 in partnership with AI4bharat. The model aims to deliver GPT-3.5-like performance on Indic languages under compute and data constraints, achieving high performance on a frugal budget.

Mendel has developed Hypercube, an AI platform that integrates Meta Llama 2-based LLM for natural language processing on clinical data. Hypercube uses a hypergraph semantic layer and the proprietary Eloquent language to enable querying and retrieving clinical information for applications like trial matching and patient cohorting. Llama 2 assists in converting natural language queries into Eloquent, bridging the gap for users unfamiliar with the query language.

Mayo Clinic uses Meta's Llama 2 to develop RadOnc-GPT, a specialized large language model for radiation oncology. RadOnc-GPT is fine-tuned on patient records and deployed locally to ensure data security. It supports a chatbot that answers routine post-radiotherapy questions, reducing nurses’ and clinicians’ workloads and improving treatment decision-making speed, accuracy, and quality. Future developments include predicting patient outcomes and expanding clinical tasks, enhancing operational efficiency and patient care.

Meditron, a suite of open-source large multimodal foundation models tailored for the medical field, leverages Meta's Llama 2 and 3 to assist with clinical decision-making and diagnosis in low-resource settings. Developed by researchers at EPFL and Yale School of Medicine in collaboration with humanitarian organizations like the International Committee of the Red Cross, Meditron provides evidence-based, contextually aware recommendations and diagnostic support. It automates tasks such as generating treatment regimens, determining radiation modalities, and assigning diagnostic codes, thereby enhancing clinical efficiency and improving patient outcomes. With over 30,000 downloads and top performance on medical benchmarks, Meditron democratizes access to advanced medical AI tools, enabling equitable healthcare innovation in underserved regions.

Elyza, a Japanese AI startup, uses Meta Llama 2 to develop a Japanese large language model designed for advanced natural language processing applications, including language translation, chatbots, and text generation, thereby enhancing AI-driven communication and language services in Japan.

Taiwan LLM is a pioneering large language model focused on Traditional Chinese used in Taiwan. Built on Meta Llama 2, it incorporates cultural context and advanced pre-training on comprehensive datasets, enabling it to understand linguistic nuances and cultural references, thus revolutionizing Traditional Chinese language processing in NLP applications.

Odia Generative AI is an open-source project aimed at expanding LLM capabilities for the Odia language, spoken by 40 million in India. Using Meta Llama 2, OdiaGenAI has developed the Odia Llama, a fine-tuned LLM for Odia, and is exploring applications like AI chatbots and AI tutors, thereby enhancing accessibility and digital inclusion for Odia speakers.

Researchers at UC Berkeley have introduced RAFT (Retrieval-Augmented Fine-Tuning) by leveraging Meta Llama 2 on Azure AI Studio. RAFT enhances domain adaptation in language models by improving their ability to retrieve and integrate relevant information. This novel approach benefits specialized applications by making Meta Llama 2 more versatile and adaptable to various domain-specific tasks.

ML Commons has integrated Meta's Llama 2 70B parameter model into version 4.0 of its MLPerf Inference benchmark. The published results demonstrate the performance potential of various platforms for running one of the most demanding and capable large language models, highlighting the efficiency and scalability of Meta’s Llama 2 in high-performance computing environments.

Pratham developed 'BaalSakhi', a chatbot powered by Meta's Llama 2 (70B), integrating via WhatsApp to provide personalized, on-demand support for early childhood care and education to parents and caregivers in resource-poor settings. BaalSakhi draws insights from trusted sources like government departments, UNICEF, WHO, and childcare experts, helping users navigate their child's early development journey independently.

Upstage fine-tuned Meta's Llama 2 to create Solar 70B, a large language model that secured the top position on the Open LLM Leaderboard upon its debut, showcasing the potential of open innovation in AI.

Mathpresso's QANDA platform utilizes Meta's Llama 2 to create MathGPT, a math-specific large language model that enables highly personalized learning experiences. By leveraging open-source models, Mathpresso offers flexible, domain-specific educational products that democratize access to quality education, allowing learners worldwide to receive tailored math support.

NoHarm.ai, a healthcare non-profit startup in Brazil, uses Meta Llama 2 and an open-source web application to improve hospital discharge summaries for patient care transitions. The NoHarm Discharge Summary tool extracts key information from inpatient records using Named Entity Recognition (NER) and generates discharge summaries in Portuguese for physicians to validate, enhancing data organization and communication within the healthcare system.