这是用户在 2024-6-8 14:08 为 https://mailchi.mp/slator/slator-weekly-jun-7-2024?e=d2acdcbee3 保存的双语快照页面,由 沉浸式翻译 提供双语支持。了解如何保存?
Copy 复制
View in Browser  在浏览器中查看

Buy or Build 购买或建造

There are a gazillion market maps for large language models and the app layers built on them. So at this point in the cycle, there’s a lot of value in the simple act of curation.
大型语言模型和基于它们构建的应用程序层有无数的市场地图。因此,在这个周期的这一点上,简单的策展行为有很大的价值。


We did just that this week. Starting with SeamlessExpressive from Meta back in the olden days of November 2023 all the way to OpenAI’s GPT-4o, released in May 2024, we listed 10 (foundational) large language models that we think matter to the language industry.
这周我们就是这样做的。从 2023 年 11 月 Meta 的 SeamlessExpressive 开始,一直到 2024 年 5 月发布的 OpenAI 的 GPT-4o,我们列出了 10 个我们认为对语言行业重要的(基础)大型语言模型。


The pace of new LLM releases is relentless. As a language AI agency or a translation and localization management platform you have to make a decision. Do you bet on a single model and risk lock-in?
新LLM版本的发布步伐是不懈的。作为语言人工智能机构或翻译和本地化管理平台,您必须做出决定。您是否押注于单一模型并承担风险锁定?

Do you strive to be model agnostic so you can plug in the latest and greatest models as they become available? Or do you build your own model, fully customized for the task at hand?
您是否努力做到与模型无关,以便可以在最新、最好的模型可用时插入它们?或者您是否构建自己的模型,针对手头的任务完全定制?


Unbabel chose the latter option. This week, the language AI agency announced the release of TowerLLM. Unbabel calls it the first LLM designed, trained, and optimized for translation and claims it outperforms all other rivals, including GPT-4o.
Unbabel 选择了后者。本周,语言 AI ​​机构宣布发布 TowerLLM。 Unbabel 称其为第一个为翻译而设计、训练和优化的LLM,并声称其性能优于所有其他竞争对手,包括 GPT-4o。


Localization platform Phrase chose option two. Asked if they are building Phrase as a model agnostic platform, CEO Georg Ell, explained on this week’s SlatorPod that [while] they “are doing most of their work on Open AI at the moment” he wanted to make sure that they “weren't creating a single vendor dependency.”
本地化平台Phrase选择了方案二。当被问及他们是否正在将 Phrase 打造为一个与模型无关的平台时,首席执行官 Georg Ell 在本周的 SlatorPod 上解释说,[虽然]他们“目前正在做开放人工智能方面的大部分工作”,但他希望确保他们“是”不要创建单一供应商依赖性。”


The agenda is now complete for SlatorCon Remote on June 26th. From an industry overview, to the new ISO Standard 5060, RAG in the Language Industry, loc lessons learned from a right-to-left (RTL) project, improving LQA, and an MT market review —we have it all.
6 月 26 日 SlatorCon Remote 的议程现已完成。从行业概览到新的 ISO 标准 5060、语言行业中的 RAG、从从右到左 (RTL) 项目中汲取的经验教训、改进 LQA 以及 MT 市场回顾,我们一应俱全。


And registrations are open for SlatorCon Silicon Valley 2024 on September 5th. Super Early Bird tickets are available now. See you in Menlo Park!
2024 年 SlatorCon 硅谷大会将于 9 月 5 日开放报名。超级早鸟票现已发售。门洛帕克见!


If you missed SlatorCon London check out our Youtube channel for video interviews with speakers and panelists.
如果您错过了 SlatorCon London,请查看我们的 Youtube 频道,观看演讲者和小组成员的视频采访。

 

10 Large Language Models That Matter to the Language Industry
对语言行业至关重要的 10 个大型语言模型

An overview of ten recent large language models, with breakthrough translation, voice AI, and speech-to-speech capabilities.
概述最近十种大型语言模型,具有突破性的翻译、语音人工智能和语音转语音功能。

The Best AI for Translation
最佳翻译人工智能

A breakthrough in LLMs for translation, this new model outperforms machine translation tools like GPT-4o, DeepL and Google.
这种新模型是翻译领域LLMs的突破,其性能优于 GPT-4o、DeepL 和 Google 等机器翻译工具。
Learn More  了解更多
Improving the Results of LLM
改善LLM的结果

The rationale for RAG, leveraging your Translation Memories and Termbases, best practices and pitfalls, plus two real use cases.
RAG 的基本原理,利用您的翻译记忆库和术语库、最佳实践和陷阱,以及两个真实用例。
Only 3 Weeks Left
仅剩 3 周时间
All Things Language at SlatorCon
SlatorCon 上的一切语言

The Language Industry Transformation, Investor Panel, Language AI Stack, The Future of Language Work, Speech Translation, Networking and more.
语言行业转型、投资者小组、语言人工智能堆栈、语言工作的未来、语音翻译、网络等。
🐥 Super Early 🐥
🐥超级早🐥
Are you happy with your TMS?
您对 TMS 满意吗?

Yes, definitely 当然是
Does the job 做这份工作
Not really, needs improvement
不太好,需要改进
Access the 2024 Market Report via 👉 our Growth, Pro, Enterprise plans
通过我们的增长、专业、企业计划访问 2024 年市场报告

Phrase CEO Georg Ell on the Arms Race in Language Technology
Phrase 首席执行官 Georg Ell 谈语言技术的军备竞赛

Georg Ell on scaling Phrase to become a platform, their recent pricing changes, the challenge of fully automating LQA, and their recent launch of Next GenMT.
Georg Ell 介绍了将 Phrase 扩展为一个平台、他们最近的定价变化、完全自动化 LQA 的挑战以及他们最近推出的 Next GenMT。

UN Launches Speech Bank for Interpreters to Help with Exam Preparation
联合国为口译员推出语音库以帮助准备考试

The UN's DGACM launched a Speech Bank to help interpreters and students practice for interpreting examinations with real speeches in six languages.
联合国 DGACM 推出了语音库,帮助口译员和学生用六种语言的真实语音练习口译考试。

Language Industry Data and News Briefing June 2024
语言行业数据和新闻简报 2024 年 6 月

Language industry’s definitive monthly news and data briefing; key M&A and funding, financial results, hires, company KPIs, product launches, and social media highlights. Subscriber Content
语言行业权威的每月新闻和数据简报;关键并购和融资、财务业绩、招聘、公司 KPI、产品发布和社交媒体亮点。订阅者内容

Slator Language Industry Buyer Tracker for June 2024
2024 年 6 月 Slator 语言行业买家追踪

In the June 2024 edition of our buyer tracker, we provide Slator subscribers with a selection of some of the most important changes among buy-side decision-makers. Subscriber Content
在我们的买家追踪器 2024 年 6 月版中,我们为 Slator 订阅者提供了买方决策者中一些最重要变化的精选内容。订阅者内容

Translating Millions of Words per Week Is Just the Start for Meta
每周翻译数百万字只是 Meta 的开始

At SlatorCon London 2024, Iris Orriss, VP of Global Experience and International Marketing, discusses why AI translation still needs humans to create “truly local” content.
在 2024 年伦敦 SlatorCon 大会上,全球体验和国际营销副总裁 Iris Orriss 讨论了为什么人工智能翻译仍然需要人类来创建“真正本地化”的内容。

Here’s a New Dataset for Emotion-Aware Speech Translation
这是用于情感感知语音翻译的新数据集

Researchers from academia and industry present MELD-ST dataset, a new resource designed to advance emotion-aware speech translation.
来自学术界和工业界的研究人员提出了 MELD-ST 数据集,这是一种旨在推进情感感知语音翻译的新资源。

Language Service Providers Are Moving into Multilingual AI Text Generation
语言服务提供商正在转向多语言人工智能文本生成

A 2024 Slator survey reveals how popular multilingual AI text generation as a service is becoming among language service providers.
2024 年 Slator 的一项调查揭示了多语言人工智能文本生成服务在语言服务提供商中的流行程度。

How to Get First-Rate Localization for Second-hand e-Commerce Retailer Vinted
二手电商零售商Vinted如何获得一流的本地化

Localization Director Martina Pancot walks through her approach to establish, and to continue to improve, a localization program for the platform at SlatorCon London 2024.
本地化总监 Martina Pancot 在 2024 年 SlatorCon London 上介绍了她为该平台建立并继续改进本地化计划的方法。

AppTek, Deluxe, and EBC on AI-Enabled Localization in Public Broadcasting
AppTek、Deluxe 和 EBC 探讨公共广播中的人工智能本地化

AppTek, Deluxe, and EBC discuss the advancements and remaining challenges of AI dubbing, including content types, languages, audiences, and quality expectations.
AppTek、Deluxe 和 EBC 讨论了人工智能配音的进步和剩余挑战,包括内容类型、语言、受众和质量期望。

Tencent AI Introduces an LLM-Based Virtual LSP for Literary Translation
腾讯AI推出基于LLM的文学翻译虚拟LSP

Researchers from Tencent AI Lab, Monash University, and the University of Macau present a multi-agent virtual LSP based on large language models and specialized in literary translation.
来自腾讯人工智能实验室、莫纳什大学和澳门大学的研究人员提出了一种基于大语言模型、专门从事文学翻译的多智能体虚拟LSP。

UK Health Service NHS Announces Potential USD 93m Language Services Contract
英国医疗服务 NHS 宣布潜在价值 9300 万美元的语言服务合同

The UK’s NHS published a notice for the potential procurement of language services worth about USD 93m (GBP 73m), including a one-stop-shop contract and a T&I app. Subscriber Content
英国 NHS 发布了一份可能采购价值约 9300 万美元(7300 万英镑)语言服务的通知,其中包括一站式合同和 T&I 应用程序。订阅者内容

How Linguamatics Sees Language Solutions for the Life Sciences Industry
Linguamatics 如何看待生命科学行业的语言解决方案

At SlatorCon London 2024, Richard Parnell, General Manager of IQVIA’s language solutions division Linguamatics, discusses the company’s vision for Life Sciences translations.
在 2024 年伦敦 SlatorCon 大会上,IQVIA 语言解决方案部门 Linguamatics 总经理 Richard Parnell 讨论了该公司对生命科学翻译的愿景。

Straker Says FY24 Results Impacted by ‘Tougher than Expected Market Conditions’
Straker 表示 2024 财年业绩受到“市场状况比预期严峻”的影响

The New Zealand LSP confirms name change, and announces further investment in technology integrations and solutions to address the demand for generative AI solutions. Subscriber Content
新西兰 LSP 确认更名,并宣布进一步投资技术集成和解决方案,以满足生成式人工智能解决方案的需求。订阅者内容

Two European Audiovisual Translation Organizations Are Teaming Up
两个欧洲视听翻译组织正在联手

The European Association for Studies in Screen Translation (ESIST) and SUBTLE - the Subtitlers’ Association promote quality standards, industry networking, and education.
欧洲屏幕翻译研究协会 (ESIST) 和 SUBTLE(字幕制作者协会)促进质量标准、行业网络和教育。

Press Releases 新闻稿


🗞️ Unbabel Launches TowerLLM, Offering Best-in-Market Machine Translation With GenAI

🗞️ Phrase Advances Hyperautomation with Next Wave of AI-Powered Localization Enhancements

🗞️  Véronique Özkaya Joins Summa Linguae Technologies as Co-CEO
🗞️ Unbabel 推出 TowerLLM,通过 GenAI 提供市场最佳的机器翻译 🗞️ Phrase 通过下一波人工智能驱动的本地化增强功能推进超级自动化 🗞️ Véronique Özkaya 加入 Summa Linguae Technologies 担任联合首席执行官

Events 活动


📅  Join Elia for ND 2024 in Berlin from Oct 1-2. Early bird registrations are now open so secure your spot!
📅 与 Elia 一起参加 10 月 1 日至 2 日在柏林举行的 ND 2024。早鸟登记现已开放,请确保您的席位!

Latest Slator Research 最新的斯拉特研究


📚  *NEW* Slator 2024 Language Industry Market Report

📚  Slator Pro Guide: ISO Certifications for LSPs in the Era of AI

📚  Slator 2024 Life Sciences and Language AI Report

📚  Slator 2023 Language Industry M&A and Funding Report

📚  Slator Pro Guide: Media Localization into English

📚 Slator 2024 Interpreting Technology and AI Report

📚 Slator Pro Guide: Solution Sales for Language Service Providers

📚 Slator Pro Guide: Scaling an LSP Key Account

📚 Slator Pro Guide: Translation AI
📚 *新* Slator 2024 年语言行业市场报告 📚 Slator Pro 指南:人工智能时代 LSP 的 ISO 认证 📚 Slator 2024 年生命科学和语言人工智能报告 📚 Slator 2023 年语言行业并购和融资报告 📚 Slator Pro 指南:媒体本地化英语 📚 Slator 2024 年口译技术和人工智能报告 📚 Slator Pro 指南:语言服务提供商的解决方案销售 📚 Slator Pro 指南:扩展 LSP 关键客户 📚 Slator Pro 指南:翻译 AI

Copyright © 2024 Slator, All rights reserved.
版权所有 © 2024 斯拉特,保留所有权利。



Want to change how you receive these emails?
想要更改接收这些电子邮件的方式吗?

You can update your preferences or unsubscribe from this list
您可以更新您的偏好或取消订阅此列表

 
set 限制解除
×