Hey there,
In this newsletter, we’re spotlighting our new academic program featuring the first four participating universities, unveiling our latest benchmark study on GenAI for translation, walking you through the new features and LLMs on our GenAI portal, and sharing some team updates.
University Partnerships
We recently kicked off our university partnership program to aid students studying translation and localization using Intento tools. Four universities have already joined the program:
- The Middlebury Institute of International Studies, USA
- New York University, USA
- Università di Padova, Italy
- Université de Montréal, Canada
Let us know if your alma mater wants to join, too. Together, we aim to contribute to research and development and support the future workforce in localization!
Generative AI for Translation in 2024
Our recent blog post explores the new generative AI models released in late 2023 and their translation features. Looking at LLMs from Anthropic, Google, Meta, and OpenAI, we tested their translation quality, performance, and cost against more traditional MT engines.
Another of our blog posts explores GenAI-based summarization in different languages and how it can be enhanced using traditional machine translation.
Also, if you haven’t had time to follow all the news in the GenAI world, we have prepared some important highlights of what happened during the last three months.
New features
More control over API key rotation
Stay secure by updating your API keys whenever you need to via Intento Console.
Powerful new models added
With the ever-evolving landscape of AI, we’re excited to introduce the latest models to the GenAI portal: GPT-4 Turbo (gpt-4-0125-preview), GPT-3.5 Turbo (gpt-3.5-turbo-0125), PaLM 2 (chat-bison-002 including the version with a 32k context size), and Claude-2.1! Gemini and Claude 3 are to come! Stay at the forefront of generative AI, and try out our GenAI Portal if you haven’t yet.
Multiple parameters for language skills
Our GenAI Portal automates language skills beyond just translation by storing and reusing prompts. The latest update introduces skill parameters, giving you more ways to define the context. For instance, you can now set up a skill to translate a message into a specific language, with options for tone of voice and gender available in dropdown menus.
Pin your skills for easy access
Pin your favorite and most-used language skills in the GenAI Portal for easy access. Now, your go-to skills are just a click away.
Team updates
Meet our New VP of Language Experience
At Intento, we think the goal of localization is to create a seamless and genuine language experience throughout the entire customer journey, from initial website engagement to real-time customer support chats. To accomplish this for our customers, we develop tailored solutions using the Intento Language Hub. This involves understanding customer needs and preferences, optimizing AI models, building automated translation workflows, and integrating them with the software systems your translators, customers, and employees use.
To make this process more efficient, we’ve consolidated our professional services, AI curation, and implementation teams under the leadership of our very own Sergei Polikarpov, now VP of Language Experience! Since joining Intento in 2019, Sergei has led our AI and Machine Translation research teams and key projects, including our annual State of Machine Translation report.
New VP of Customer Success: Katya
We’re excited to introduce Katya, Intento’s new VP of Customer Success!
With her proven track record in elevating customer experiences for top brands like Wrike and Veeam, we’re confident that her leadership in customer success will further strengthen our commitment to delivering Intento’s value to our customers.
Welcome aboard, Katya.
The latest in artificial intelligence
- Anthropic just announced the next generation of Claude, the Claude 3 family of models with vision capabilities. Opus, the most intelligent model, outperforms its peers on most of the common evaluation benchmarks for AI systems.
- Google announced the new Gemini 1.5 with an astonishing 1M context size (still in a limited preview) and an open LLM called Gemma.
- Mistral just launched their Large model with native support of English, French, Spanish, German, and Italian.
- Meta’s Llama 3 is now planned to be released in July 2024.
- Microsoft Research and University of Chinese Academy of Sciences proposed a 1-bit LLM variant, namely BitNet b1.58 that matches the full-precision Transformer LLM and paves the way to significantly more cost-effective training and inference.
- NVIDIA introduced Nemotron-4 15B, a 15-billion-parameter multilingual LLM. It exhibits the best multilingual capabilities of all similarly sized models, even outperforming models over four times larger and those explicitly specialized for multilingual tasks.
- A group of researchers from Allen Institute for Artificial Intelligence and several universities published the first release of OLMo, a state-of-the-art, truly Open Language Model and its framework to build and study the science of language modeling.
- Silo AI together with the University of Turku and HPLT successfully completed training the Poro model. Poro is designed to process English and Finnish and has proficiency with a variety of programming languages. Additionally, it can perform translation between English and Finnish.
- How large context LLMs could change the status quo.
