Post-Training Is How You Keep Your Taste | Fireworks CEO Lin Qiao

Okay, up next we have Lynn, uh close close friend and collaborator of Brendan's. Um Lynn, uh actually show of hands, who here who here does post training?

接下来有请 Lynn。她是 Brendan 的好友和合作者。Lynn，请先调查一下，在座有谁从事后训练？

Okay, good amount of the room. And who here uses Firework? Okay, good amount of the room, too. So, Lynn, you have you have some friendlies in the audience.

在座有不少人。那么，有谁使用 Firework？看来也有不少人。Lynn，观众中有一些你的熟人。

Um and so for this next talk, what we're going to do is we're going to focus on post training.

接下来的演讲将重点讨论后训练。

Um Lynn, si- similar setup to to Brendan's talk. We're going to do 15 minutes on how to approach the problem of post training, um and then 15 minutes or so for Q&A. Uh and Lynn, really delighted to have you here. Thank you for joining us.

Lynn，本场演讲的形式与 Brendan 的演讲类似。我们将用 15 分钟讲解如何处理后训练问题，再用 15 分钟左右进行问答。Lynn，很高兴你能来到这里，感谢你的参与。

Thanks for having me. Uh hi everyone, good morning. I'm Lynn, I'm CEO and co-founder of Firework. So, as we bring up the slide, today I'm going to do a little bit uh deep dive on post training and why post training could be very relevant to uh you building your own business. So, first of all, a little bit like prior, I think um this year we have seen So, first of all, I'm a Firework weather specializing intelligence platform.

感谢大家邀请我。大家早上好，我是 Lynn，Firework 的首席执行官兼联合创始人。在加载内容的同时，我想深入探讨一下后训练，以及后训练为什么可能与你打造自己的业务高度相关。首先，与前面提到的内容类似，我认为今年我们已经看到……在此先介绍一下，Firework 是一家专注于专业化智能的平台。

There are tons and tons of application built on top of us, all the way from startups to digital native and enterprises. So, we get uh um the fun part of my job is we get to see a lot of patterns.

我们的平台上构建了大量应用，客户涵盖初创企业、数字原生企业和大型企业。因此，我的工作中很有意思的一点是，我们能够观察到很多规律。

Um what are the innovation people are uh building on top of us and uh and how they are what are challenges that are uh they're facing and what are trend um people being developers are um building on top of us. So, uh one of the things, especially in the past 1 year, software development and application development has been somewhat disrupted because it goes fine in the past, um you have good idea, you want to implement

例如，人们正在基于我们的平台开展哪些创新、面临哪些挑战，以及开发者基于我们的平台构建产品时呈现出哪些趋势。尤其是在过去 1 year，软件开发和应用开发都在某种程度上被颠覆了。过去，如果你有一个好想法，并希望将它

it, scaling production, it requires a team of tens of very strong product engineers, PMs working together, multiple quarters to deliver that. And right now, with one person, a few weeks, uh without understanding how to write a single line of code, you can do that.

实现并扩展到生产环境，就需要一支由数十名优秀的产品工程师和产品经理组成的团队协同工作，耗费数个季度才能交付。如今，一个人只需几周就能做到，即使他完全不懂得如何编写一行代码。

So, that collapsing of resource required both in terms of timeline and deep expertise is shifting how how competitive the application space is, and it's shifting people from thinking about building on top of off-the-shelf off-the-shelf black box API to build a much deeper mode. So, that they can build a much durable business.

无论在时间周期还是深度专业知识方面，所需资源都已大幅减少。这正在改变应用领域的竞争格局，也促使人们不再考虑基于现成的黑盒 API 进行构建，而是转向更深层次的模式。通过这种方式，他们能够打造更持久的业务。

As also you have seen lot of discussion, especially in the past 1 week, between open model and the closed model and all the rallying and support across the open model.

大家也看到了许多相关讨论，尤其是在过去 1 week，讨论主要围绕开放模型和封闭模型展开，其中还包括各方对开放模型的声援与支持。

The depth of that that alliance is because we believe I mean the industry is much deeper because look at the whole entire industry, there are so many companies, right? There are so many company, all of you are building your own company.

这种支持之所以如此坚定，是因为我们相信这个行业具有深刻得多的内涵。纵观整个行业，存在着许多公司，在座各位也都在打造自己的公司。

Every single company exists for a reason because they focus on solving a unique problem in a special way.

每家公司都有其存在的理由，因为它们都专注于以独特的方式解决特定问题。

And that means they carry their own judgment, taste, and determination, conviction into that product, and that's why company exists.

这意味着，它们会将自己的判断和品味，以及决心和信念注入产品。这正是公司存在的意义。

Today, if you build on top of off-the-shelf API, then you really need to think about how you keep that special taste judgment and unique part forward. And we believe one approach for every company to build a build durable business is to actually bake your judgment, taste, and customer deep understanding into the intelligence you build on top of instead of just a off-the-shelf API. So, that's kind of a

如今，如果你基于现成的 API 进行构建，就必须认真思考如何继续彰显独特的品味和判断，并保持自身的差异化。我们相信，每家公司要想打造持久的业务，一种做法就是将自己的判断、品味以及对客户的深入理解真正融入其所依托的智能，而不是仅仅依赖现成的 API。以上是

little bit context of what's happening in industry, what we are seeing, and why post training could be very very relevant to you.

一些背景，介绍了行业正在发生的变化、我们观察到的情况，以及后训练为什么可能与你们息息相关。

Okay. So, you probably heard a lot about owning intelligence not rent.

大家可能经常听到一种说法：要拥有智能，而不是租用智能。

And what does owning your own intelligence mean? It actually means many things. So, first of all, it start from data.

那么，拥有自己的智能意味着什么？它包含许多方面。首先，智能始于数据。

Intelligence is derivative of data.

智能源自数据。

And obviously the the foundation labs, the all the foundation model we're using are building on top of the public data and the label data that is has solved common tasks, but all of you are solving a specific task. That's why you you're building a business, you're building a company, and be able to curate production data with high quality and even generate synthetic data to enrich your production data is one

各家基础模型实验室以及我们使用的所有基础模型，显然都是基于公共数据和标注数据构建的。这些数据解决的是通用任务，但你要解决的是一项具体任务。这正是你创业并创办公司的原因。你可以高质量地整理生产数据，甚至生成合成数据来丰富生产数据。这是

step of owning your own intelligence.

拥有自有智能的第一步。

And then after you have data, you will start to kind of use that data turn into a model, build on top of existing model and own the weights. And there are a collection of techniques you can use to get there.

有了数据之后，你会开始利用这些数据打造一个模型，在现有模型的基础上进行构建，并拥有相应的权重。为实现这一点，有一系列技术可供使用。

Those techniques are tailored to solve different kind of problem you could possibly have. And those techniques can also interoperate with each other for you to build a reach your final goal.

这些技术分别适用于解决你可能遇到的不同类型的问题，也可以相互配合，帮助你实现最终目标。

And then after you have a great model belong to yourself solving your specific problem really well, and then you need to work on serving it.

当你拥有了一个出色的自有模型，并且它能够很好地解决你的具体问题之后，接下来就需要着手部署模型并提供模型服务。

You probably first will do a some AB testing to make sure it really move the needle for your product metrics, and then goes back in in this loop.

首先，你可能会开展一些 A/B 测试，以确保它确实能够改善产品的关键指标，然后再回到这个循环中。

Obviously, I don't think you should jump into post training right away.

当然，我认为你不应该一开始就立即进行后训练。

And and there are different phases you will go into.

你会经历不同的阶段，并逐步深入。

First, prompt everyone start from prompt, use the model as is. It's few shots. If quickly you can use that to test your ideas.

首先是提示词。所有人都会从提示词开始，直接使用原始模型，这就是少样本学习。你可以用它快速验证想法。

And then you use rag to ground um, the usage of AI into your own data and you can start to do a lot of contest engineering from that. Those are from minutes of interaction to hours of interaction. And then you first progress into hey, I have some data. I want to see how my data is going to reflect and make uh, the model work better with my product. So you will start to do supervised fine-tuning that will take

随后，你会使用检索增强生成，让AI应用以自己的数据为依据，并在此基础上开展大量上下文工程。这个过程的交互时间从几分钟到几小时不等。接下来，你已经积累了一些数据，希望了解这些数据如何发挥作用，使模型更好地服务于自己的产品，于是便会开始进行监督微调。

you a few hours to um, hey, um, the the model I want to reflect in personalized taste. It is it is very unique uh, choice of my products. So so therefore you want to start to use preferences uh, information where you collect from user interaction and uh, whether thumbs-up, thumbs-down, a lot of those kind of information and uh, help the model learn your product taste. Um, and and finally, you want to build a model towards

监督微调会花费几个小时。再往后，你会希望模型体现个性化偏好，因为这是自己产品中非常独特的选择。因此，你会开始使用来自用户交互的偏好信息，例如点赞、点踩等，让模型学习你的产品品味。最后，你会希望打造一个模型，使其具备更进一步的能力。

understand carrying your expertise in that domain. Whether that expertise is across uh, legal, finance, healthcare, customer support, recruiting, marketing, sales, you name it. Even in one industry vertical, there are so many subdomains.

这个模型能够理解并承载你在特定领域的专业知识。无论这些专业知识涉及法律、金融、医疗、客户支持、招聘、营销还是销售等领域，情况都是如此。即使在同一个垂直行业中，也存在许多细分领域。

So all of that is unique and special towards the product you're building. So usually um, you probably heard a lot of reinforcement learning and that is to to actually build towards a specialty.

所有这些都是你正在打造的产品所独有的。因此，你可能经常听到强化学习，它的目的就是让模型形成某方面的专长。

Um, so this progression is very similar to how we human being learn knowledge over time. Uh, for example, uh, we actually run uh, learn a lot of knowledge by reading uh, reading literature, right? In the literature it will say, hey, what is correct, what is it not correct? So this is a very similar to supervised fine-tuning. Um, and uh, as we grow, we develop our own taste and judgment of how we want to

这一过程与人类逐步学习知识的方式非常相似。例如，我们会通过阅读文献学习大量知识，文献会告诉我们什么是正确的、什么是错误的。这与监督微调非常相似。随着成长，我们会形成自己的品味和判断，并决定自己希望以何种具体方式处理问题。

conduct the specific way of we want to approach a problem, and that is preference or DPO.

这种处理问题的方式就是偏好调优或DPO。

Um and over course of time, we learn to be a doing a really good job at certain area, whether um hey, I want to be accounting accountant, and I would really know how to kind of um build into the financial financial data, or I want to be a specific set of um I want to be a dentist, and you learn the details of how to operate um with dentistry. So, all this is very similar exactly um to how we human being

随着时间推移，我们会逐渐擅长某些领域。例如，如果我想成为一名会计师，就要真正掌握如何深入分析财务数据；如果我想专攻某个方向，例如成为一名牙医，就要学习牙科诊疗的各种细节。

acquire knowledges.

这些过程都与人类获取知识的方式非常相似。

So, um and also very interesting different techniques are there to solve different kind of problems.

另一个有意思的现象是，不同的技术适合解决不同类型的问题。

Um for example, if the model doesn't know um the fact, and the fact actually is very dynamic, the facts of your data is uh in in your product is very dynamic, and then typically you use rag to solve that problem.

例如，如果模型不了解某项事实，而这项事实变化得非常快，即产品中的数据具有很强的动态性，通常就会使用RAG来解决这个问题。

Um however, if your model output all the behavior uh or um or the structure is off, then you you give you curate data and do supervised fine-tuning to correct that.

但如果模型输出的整体行为或结构不正确，就需要整理数据，并通过监督微调加以纠正。

Um if your model's answer, the quality is is personal or kind of is specific to the taste of your product, um and then you use preference and tuning.

如果模型答案的质量标准具有个性化特点，或需要满足特定需求并体现你的产品调性，就需要进行偏好调优。

Or the model is quite weak on the special problem you're trying to solve, then you use RL uh reinforcement learning.

如果模型在你要解决的特定问题上表现很弱，就应采用RL，也就是强化学习。

And the end result of the model is too slow or too expensive to for you to serve in production, and then you use distillation to um go um let the teacher teach the student much smaller student model, so it can be more much more performing or economical.

如果最终得到的模型速度太慢或成本太高，无法部署到生产环境中，就可以采用蒸馏，让教师模型教导规模小得多的学生模型，从而使其性能高得多，或者成本更低。

Obviously, distillation also means different things for LLM, VLM, or uh image generation models. Um if you are interested, we can talk more about that.

当然，对于大语言模型、视觉语言模型或图像生成模型，蒸馏也有不同的含义。如果大家感兴趣，我们可以进一步探讨。

So, there are many different ways um teams start to hey, I want to kind of try those technology, and um and they may not be happy uh of the experience because they can burn money um and the time, but may not reach their uh ideal results. So, so here are the areas that you can possibly uh feel frustrated. Uh for example, um when it comes to data, and data is the essence of tuning, it the quantity is

因此，团队可以通过许多不同的方式着手尝试这些技术，但也可能对整体体验感到不满，因为他们投入了大量资金和时间，却未必能获得理想的结果。以下几个环节可能会令人感到沮丧。首先是数据，它是调优的核心。

not the most important. Actually, quality is the most important. But, sometimes just by throwing uh tons and tons of data into a training process may not leading to a great outcome. So, you really need to control the data quality and have And usually, who's the best judge of data quality? It's actually your product team. Um so, that's where we see the convergence of um before GenAI, product team and the research team or ML

在数据问题上，数据量并不是最重要的，真正重要的是质量。有时，仅仅将海量数据投入训练过程，未必能够带来理想的结果。因此，必须把控数据质量。通常，最有资格判断数据质量的是产品团队。由此，我们看到不同团队正在融合。在生成式AI出现之前，产品团队与研究团队或机器学习团队通常各自独立。

team, they're separate organizations.

这些团队分属不同的组织。

They're separate team. And they work hand-in-hand to make things happen. And a lot of time these days, when um people post-train uh GenAI models, we see kind of the convergence of the product team needs to make judgment call of the data quality and start to be deeply involved in the process. So, uh they can ensure uh the best result.

尽管这些团队彼此独立，但会携手合作，共同完成任务。如今，当人们对生成式AI模型进行后训练时，团队往往开始融合：产品团队需要判断数据质量，并深度参与整个过程，从而确保获得最佳结果。

Um and the second is you going to have evals. Um I know all of you are very busy, and uh part into launch quickly, and a lot of evals is a vibe evaling.

第二点是需要进行评估。大家都很忙，也希望尽快发布产品，因此许多评估只是凭感觉进行。

[laughter] And the founders uh really look at the result and feel hey, it is it right or not? Actually, this is judgment. This is you put your judgment of uh of end result and decide whether it's good or not. And that judgment you convert into systemic evaluation. So, this is no different from traditional software development where you have your unit test, integration test to ensure quality. Similarly,

（笑声）创始人会认真审视结果，思考这个结果究竟是否正确。这其实就是判断，即对最终结果作出判断，决定它是好是坏。接下来，还要将这种判断转化为系统化的评估。这与传统软件开发并无不同。在传统软件开发中，需要进行单元测试和集成测试，以确保质量。

if you want to if you think about doing post-training and then have a way to build the eval and build your own judgment into a repeatable process, it's extremely important.

同样，如果想进行后训练，就需要设法建立评估体系，将自己的判断融入一套可重复执行的流程。这一点极其重要。

Uh and then there could be when you do RL, there could be sloppy RL environment uh where you build a simulation and the simulation is not really reflecting the reality and then the model can heel climb on on kind of the a bad um simulation and I mean it it it will it will also could possibly do reward hacking um and all kind of weird stuff. So, a fun story about reward hacking is um we have been asking a model to uh to

此外，在进行强化学习时，强化学习环境可能会十分粗糙。例如，你搭建了一个模拟环境，但它无法真正反映现实，模型就可能在这种糟糕的模拟环境中不断爬坡优化，甚至可能进行奖励作弊，做出各种奇怪的事情。关于奖励作弊，有一个很有意思的故事：我们曾要求一个模型

generate to do this is coding example uh to create um to to generate code that minimize uh the the error compilation error.

生成代码。这是一个编程示例，要求模型生成尽可能减少编译错误的代码。

Okay. So, guess what the model did? The model generate zero line of code.

猜猜模型是怎么做的？它生成了零行代码。

Okay, there's no compilation error, but that's absolutely not what you want. So, this is a kind of one example of reward hacking, it's very common because model was very smart, it'll try in all different way to uh to get um your goal, but it may not be what you want. So, uh pay attention to all these details and uh and try not to let the model outsmart you.

这样确实不会出现编译错误，但显然并非你想要的结果。这就是奖励作弊的一个例子。这种情况很常见，因为模型非常聪明，会尝试通过各种方法实现你设定的目标，但结果未必是你真正想要的。因此，需要留意所有这些细节，尽量避免让模型钻空子。

Um obviously, between um your experimentation, so think about um your development process as hey, you're actually doing a lot of experimentation from training the model to uh training the model is not the end of of your experiment.

在实验过程中，可以这样理解开发流程：从训练模型开始，你始终在进行大量实验，而训练完成并不意味着实验结束。

The judge is the final judge is whether product matches is moving or not. So, you need to bring the final model into into serving tier and do AB testing. Um and that transition is extremely important because um the quality can drop if you move from one training stack to a serving stack without aligning these two. Because think about the model is tons and tons of calculation, math, and the matrix multiplication. Um and

最终的检验标准是产品指标是否有所提升。因此，你需要将最终模型部署到服务层，并进行A/B测试。这一转换极其重要，因为如果从训练技术栈迁移到服务技术栈时没有实现两者对齐，模型质量就可能下降。模型涉及海量计算、数学运算和矩阵乘法，而这些运算的具体执行方式至关重要。

the way to do multi matrix multiplication and calculation, if you use different library, different numerics, um and different optimization, it will lead different results. And therefore, the the end result of a training may not be replicated or even uh you lose precision during serving. So, that alignment is very important. I can give you more examples of that.

在执行多矩阵乘法和其他计算时，如果使用不同的库、数值精度和优化方法，就会产生不同的结果。因此，最终的训练结果可能无法复现，甚至可能在服务过程中损失精度。这种对齐非常重要。关于这一点，我还可以再举一些例子。

Um and there are a few other um challenges you will run into and happy to talk with you more about in details offline.

此外，你还会遇到其他挑战，我也很乐意在线下与大家进一步详细交流。

So, there have been many pioneers uh we worked with post-training. I would say Cursor is one of the few. They have started onboarding getting onto this train from the beginning of last year.

后训练领域已经涌现出许多先行者，我们也曾与他们合作。Cursor就是其中之一。他们从去年年初便开始踏上这段旅程。

Uh there are multiple reasons. One is they really want to control their destiny of the supply of the model. And obviously, um they have a lot of they have a lot of user engagement. They understand um how what is customer preferences, a lot of data. So, that becomes the beginning of their journey. And they do um pretty deep mid-training to post-training. And you have heard them announce um continuously newer models every few

原因有很多。其一，他们确实希望掌控自己的命运，确保模型供应自主可控。此外，他们拥有大量用户互动数据，既了解客户的偏好，也掌握着丰富的数据。这些条件成为了他们踏上这段旅程的起点。他们进行了相当深入的中期训练和后训练。大家也都注意到，他们每隔几个月就会发布更新。

month. Um and uh and the result is great. Um their inspiration is to compete at the frontier quality. Uh very bold aspiration. And they're getting there. So, very impressive result from composer two, composer 2.5, um on par with um They're always trying to be on par or beat um the frontier labs quality at the same time. So, um they are kind of one of the examples, a very vibrant example, is

这些模型取得了非常出色的成果。他们的愿景是在质量上与前沿模型竞争。这是一个非常大胆的愿景，而他们正在逐步将其实现。Composer 2、Composer 2.5的成果十分亮眼，性能足以媲美前沿模型。他们始终致力于达到前沿实验室的模型质量，甚至实现超越，同时兼顾其他方面。因此，他们是一个非常鲜活的例子，证明了这件事具有可行性。

completely doable as long as you stay focused and have the right tool and cursive build on top of us.

只要保持专注并拥有合适的工具，这件事就完全可行，而Cursor正是基于我们的平台构建的。

Um there's another there huge huge variety of examples from healthcare. For example, Doximity is one example where they do clinic AI where they basically let doctors ask deep medical questions matching symptoms to medication and side effects and have well-rounded research around everything in medical space.

规模庞大的医疗健康领域也有各种各样的案例。例如，Doximity就是其中之一。他们开发了临床AI，可以让医生提出深入的医学问题，将症状与药物及其副作用相匹配，并围绕相关问题开展全面研究，覆盖整个医疗领域。

They actually also train on Fireworks and they top a very important benchmark which is Stanford Harvard clinic safety benchmark.

他们也在Fireworks上训练模型，并在一项非常重要的基准测试——Stanford-Harvard临床安全基准测试中名列前茅。

And we are very proud of that result and they they keep working on this training loop. Um Factory, that's another Sequoia company.

我们对这一结果感到非常自豪，他们也仍在不断迭代这一训练闭环。Factory则是另一家由Sequoia投资的公司。

They build on top of Fireworks and especially focus on security part of the coding.

他们基于Fireworks进行开发，尤其专注于解决编程中的安全问题。

That is a very hard topic because security is not high tolerance. You need to get it right.

这是一个非常棘手的课题，因为安全问题的容错率很低，必须确保准确无误。

They tune a model that also top the kind of benchmarking security area.

他们调优后的模型也在安全领域的相关基准测试中名列前茅。

So you can see all these examples is a demonstration. They solve all the companies are solving a very unique problem in a special way and they have been able to kind of build their own challenges use their own data by building on top of open model.

这些案例共同说明：这些公司都在以独特的方式解决各自特有的问题，并且已经能够逐步构建自身能力、利用自有数据，以及基于开放模型进行开发。

And those are the details. Obviously, as you know, last year 2025 is the year of coding. There Fireworks we support all the coding companies build on top of us. A lot of success in mid-train to post-train very strong models.

以上就是具体情况。正如大家所知，2025年是编程之年。Fireworks为所有基于我们平台开发的编程公司提供支持，并在从中期训练到后训练的整个过程中取得了巨大成功，训练出非常强大的模型。

As you can see those benchmarks are very impressive and all these coding companies are inspired to get on par or beat Frontier Labs in the coding space, but it's not coding. And we start This is the year we start to see very interesting development in all different kind of co-work space from general purpose co-work to specialized domain specific co-work as I mentioned.

这些基准测试的成绩十分亮眼。所有这些编程公司都希望在编程领域比肩甚至超越前沿实验室，但后训练的应用并不局限于编程。今年，我们开始看到各种不同类型的协作空间出现非常有趣的发展，涵盖从通用协作空间到我刚才提到的专业垂直领域协作空间。

There are like legal, finance, marketing, recruiting, sales, customer support co-work space. They are all starting to post train and own their own intelligence power their product.

例如，法律、金融、营销、招聘、销售和客户支持等协作空间都开始开展后训练，打造自有智能，为各自的产品提供动力。

Here, Jen Spark is one of the generic co-work application and they build deep research for professionals and slide generation.

Genspark是一款通用协作应用，可以为专业人士提供深度研究和幻灯片生成功能。

As you see, this they compete with a frontier model and it's slightly better, but the cost is significantly lower. So here we're talking about five to 10 10 times cost reduction. So as a startup, you can once you hit a problem like if it you can scale quickly into a doable business. And I guess another health care example and health care data or use case is not well captured

从这里可以看到，它不仅能够与前沿模型竞争，效果甚至略胜一筹，而且成本低得多，降幅可达5到10倍。因此，对于初创公司而言，一旦实现产品市场契合度，就能迅速扩大规模，发展成一家可持续经营的企业。医疗领域还有另一个例子：医疗数据或用例并未得到充分涵盖。

in in frontier labs model. So you can see the quality of tuned model is significantly better than the state of art closed model.

前沿实验室的模型对这些医疗数据或用例覆盖不足。因此可以看到，调优后模型的质量明显优于目前最先进的封闭模型。

Well, we have many different type of developers build on top of fireworks to do post training. So what what are these types?

许多不同类型的开发者都在基于Firework进行后训练。那么，他们分别属于哪些类型？

So we have seen many frontier agent builder.

我们见过许多前沿智能体开发者。

So frontier agent builder, they usually build bespoke customized harness.

前沿智能体开发者通常会构建高度定制化的框架。

They don't use common harnesses. And to co-optimize their harness with their systems and all different kind of tools they want to build on top of often time requires post training. Uh so, we have we have seen a lot of repeatable success on doing that.

他们并不使用通用工具链。要使其工具链与自身系统及计划在其上构建的各种工具实现协同优化，通常需要进行后训练。我们已经看到许多可复制的成功实践。

And obviously, we have seen a lot of big companies put to post training because interestingly, those incumbents have huge amount of traffic.

我们也看到许多大公司投入后训练，因为这些行业巨头拥有巨大的流量。

For them to deploy AI features across the board to all the all their customers is a huge cost burden. Huge cost burden. Uh so, if we say we we need to be careful not scale into bankruptcy, it's not just for startups. Uh it's actually for incumbents. They literally their CFO is blocking their AI feature launch because of the cost. And the post training is way to remediate that.

对这些公司而言，部署AI功能并将其全面覆盖至所有客户，会带来极其沉重的成本负担。因此，“必须谨慎，不能因扩张而破产”这一原则不仅适用于初创公司，也适用于行业巨头。这些公司的CFO确实会因成本问题阻止AI功能上线，而后训练是解决这一问题的一种方式。

Uh and we have seen a lot of specialized model operator, uh those could be uh new uh new labs and a lot of kind of cutting-edge model develops also um do significant post training.

我们还看到许多专业的模型运营商开展大量后训练，其中可能包括新兴实验室；许多前沿模型开发者也在这样做。

So, uh this is a quote we have heard repeatedly uh that after part of market fit, post training becomes the vehicle to for many, many companies to build specialized intelligence. We do believe we do see the trend that in the future there could be millions of specialized models. One application per use case. It it it it will be millions. That's how uh I believe in that.

我们经常听到一种反复出现的说法：在实现产品市场契合之后，后训练就成为许多公司打造专有智能的途径。我们确实看到了这一趋势：未来可能会出现数百万个专用模型，每种用例对应一个应用，模型数量将达到数百万。

Um And obviously, there are um we have seen kind of different way to convert the data into into model. Um and the the signal from the reward and how to build those reward function is very important um part of the story.

我们已经看到多种将数据转化为模型的方式。其中，奖励提供的信号以及奖励函数的构建方式，是非常重要的一环。

Um so, our observation is no, start early, start early, and the start to do experimentation a lot more iteratively and they get hands-on experience and we we have seen people on board so quickly.

我们的建议是尽早开始，以更高频率的迭代方式开展实验，在持续迭代中积累实践经验。我们看到参与者通常都能很快上手。

There there's actually not the barrier as many people feel because this this whole reward engineering is very similar to software engineering.

实际门槛并没有许多人想象的那么高，因为奖励工程与软件工程非常相似。

The the the logical reasoning and mindset is very very similar. So so yeah so get your hands dirty and and start to kind of test it out.

两者采用的逻辑推理方式和思维模式都非常相似。因此，应当亲自上手并开始尝试。

Cool.

很好。

I think Yeah so those are the kind of key takeaways and uh I think we [snorts] we we want to acknowledge there are different specialties or different knowledges you have right now in this space. For example, we work with the companies like Cursor and Cognition.

以上是几个关键要点。此外，我们也必须承认，目前这一领域存在不同的专业方向，各方掌握的知识也不尽相同。例如，我们与Cursor和Cognition等公司合作。

They have deep researchers. They want to control every single knob as much as possible to get extreme results.

这些公司拥有资深研究人员，希望尽可能掌控每一个调节参数，以获得极致的结果。

We give them the lowest API.

我们会为这些公司提供最底层的API。

That means we just have RL rollout that they can directly interact with and they fully control the trainer.

这意味着，我们只提供可供他们直接交互的RL rollout，而训练器则完全由他们掌控。

That gives them the best result.

这种方式能够帮助他们获得最佳结果。

But many of the company you have product engineers or machine learning engineers.

但许多公司拥有的主要是产品工程师或机器学习工程师。

You have good enough information and knowledge about post training. You want to have some control but you do not want to have the lowest level control. So that's where we have our training SDK iterate very closely with all of you to get to to get you going and we also are happy to kind of deploy our own researchers. That's where we will see, um, help hands-on like training your team to to get on board. So, those are

这些工程师已经掌握了足够的后训练信息与知识，希望保留一定的控制权，但又不想深入最底层进行控制。因此，我们提供了训练SDK，可以与各位紧密协作、快速迭代，帮助各位顺利上手。此外，我们也愿意派出自己的研究人员亲自提供帮助，例如培训各位的团队，帮助团队成员上手。

the kind of different way we see different teams want to engage and and get familiar with the whole entire process.

以上就是我们观察到的不同团队参与并熟悉整个流程的几种方式。

Cool. Uh, I'm going to pause here and see if you have any questions.

我先讲到这里，看看各位是否有问题。

I'll jump um People are most most successful if they have data and reward signal already, maybe from their product. What are the what are the best examples of reward signals you've seen with very successfully trained very quickly?

我先提一个问题。如果已经拥有数据和奖励信号，例如来自自己的产品，通常最容易取得成功。你见过哪些最佳的奖励信号案例，可以让模型迅速训练成功？

So, um, usually the rewards you can think about rewards actually is code.

通常可以将奖励理解为代码。

You write rewards in code. And uh, usually think about rubrics of rewards.

奖励是通过代码编写的，通常还需要考虑其评分准则。

Um, and think about uh, you want to grade the result in multiple dimensions.

换言之，需要从多个维度对结果进行评分。

Uh, [snorts] for example, if you are building a recruiting agent, uh, and then you can think about, "Hey, how do we evaluate candidate selection?"

例如，如果你正在构建一个招聘智能体，可以思考：“我们应当如何评估候选人的筛选结果？”

Um, and a different company, I guarantee you you have different criterias. Uh, for example, we want to grade a candidate um, um, aptitude.

可以肯定的是，不同公司会采用不同的标准。例如，我们可能希望评估一位候选人的能力倾向。

Are they really hungry? They do not take no as an answer. They they will break down walls. And that's one matrix. The second is they're really fast in kind of um, building things and making progress. And and so on. So, then you have a blend of score to blend to merge this. So, so think about reward in different dimensions of rubrics. And then different company have a different way to blend those. And that's that's your

他们是否真正渴望成功？是否不达目的誓不罢休？是否会排除万难？这是其中一个维度。第二个维度是，他们构建产品、推动进展的速度是否足够快，等等。随后，你需要通过综合评分将这些维度融合起来。因此，应当从评分准则的不同维度考虑奖励设计。不同公司会以不同的方式融合这些维度。这正是你们

um, unique part and secret sauce.

独有的部分与秘诀。

Oh, sorry.

抱歉。

Uh, great talk, Lynn. Raj from Vercel.

Lynn，讲得很精彩。我是来自Vercel的Raj。

I'm curious about I think you mentioned what are the pitfalls of starting too early, but based on what you're seeing from Vercel, Vercel cognition, or other companies, when do they actually start thinking about post training? Is it when they feel ready? Is it when the cost is now too much? Because the benchmarks where they're beating it, that's a great signal that yes, you can uh um you know, beat the frontier models, but

我想请问，我记得你提到过，过早开始会带来哪些隐患。不过，根据你在Vercel、Vercel Cognition 或其他公司观察到的情况，它们究竟会在什么时候开始考虑后训练？是在认为自己已经准备就绪时，还是在成本变得过高时？从那些胜过前沿模型的基准测试来看，这无疑是一个很好的信号，表明确实可以超越前沿模型，但

is that the primary motivator? Like that is that 2% bump worth the cost or investment into this whole framework?

这是主要驱动力吗？例如，为了获得2%的提升，是否值得投入相应成本，搭建整套框架？

And and how do they think about ROI? You know, close models are more expensive, but then post training has some intercept of, you know, money and investment and maintenance going forward. So, I'm curious of like, are they approaching when cost becomes a concern or is it when the usage spiking up and they really have to now plan for a year or two in advance?

它们又如何看待投资回报率？封闭模型的成本更高，但后训练同样需要投入资金，后续还需要持续投入和维护。因此，我想知道，它们是在成本成为问题时才着手处理，还是在使用量激增、确实需要提前一两年规划时才采取行动？

Yeah, this is this is excellent question.

这是一个非常好的问题。

Uh so, usually so, think about uh in the AI building product market fit and the scaling the business as actually two phases.

通常可以这样理解：在打造AI产品时，实现产品市场契合度与扩大业务规模实际上是两个阶段。

And then in the SaaS time, it's one concept. You hit a product market fit, just scale.

而在软件即服务时代，这两个阶段属于同一个概念：一旦实现产品市场契合度，接下来只需进行扩张。

I know you got to scale as much as possible.

企业只需要尽可能扩大规模。

Um but now we see a bifurcation.

但现在，我们看到这两个阶段开始分化。

Product market fit doesn't really mean you have a durable scalable business.

实现产品市场契合度，并不真正意味着企业已经拥有一项可持续、可扩展的业务。

And typically, we see uh companies deploy the strategy of focus on product market fit first by building on top of um you know, uh Frontier Labs model because you don't need to worry about anything. So, just kind of spend your money and hit a product market fit.

通常，我们看到企业采取的策略是，先专注于实现产品市场契合度，并基于前沿实验室的模型构建产品，因为这样无需顾虑其他问题。企业只需投入资金，直至实现产品市场契合度。

Uh and the other very important thing is only after you hit the product market fit, the data you collect from product surface area are really meaningful.

另一个非常重要的方面是，只有在实现产品市场契合度之后，从产品端收集的数据才真正具有意义。

A- and you will get uh also the volume of high quality data start to collect from um product surface area. That became the fuel of uh you start to own your intelligence. So, we um we see like that as kind of the very strong indication um because once you hit upon market fit, you really think about start start to scale a business, you think about two things. One is continue to keep your competitive edge.

此时，企业也会开始从产品端收集大量高质量数据。这些数据将成为企业开始拥有自有智能的燃料。因此，我们认为这是一个非常强烈的信号。一旦实现产品市场契合度，并真正开始考虑如何扩大业务规模，企业就会考虑两件事。第一是继续保持竞争优势。

Two is um build a build durable business, so your um your revenue and your costs are, you know, in a healthy state.

第二是打造可持续发展的业务，确保收入与成本维持健康状态。

Um so, then um post-training becomes a very appealing solution because post-training allow you to um basic encode codify um your uh unique taste into into a model that no one can uh steal from. Because it's very easy to clone and copy application as is, as I as you all know, right? Uh from coding agent it's very easy. Uh so, from screenshot, boom, generate the same or

此时，后训练就成为一个非常有吸引力的解决方案，因为它能让企业将自身独特的品味编码并固化到模型中，使他人无法窃取。众所周知，原样克隆和复制应用非常容易。有了编码智能体，只需截一张图，就能生成相同甚至

even better app. Uh so, that's a very scary that application itself is kind of the a moat is is being um reduced.

更好的应用。令人担忧的是，应用本身作为护城河的作用正在减弱。

Um and you bake your data uh where reflecting the product engagement and all the deep knowledge into your model is a way to preserve that. And second is you can post-train a model that bring down the cost five to time 10 times. And and then that means you can support five to 10 times much higher traffic with the same budget. And the your unit of economics of scaling is so much better than you avoid scaling into bankruptcy.

将你的数据，即反映产品互动情况的数据，以及所有深层知识融入模型，是守住这道护城河的一种方式。第二，通过后训练模型，可以将成本降低五到10倍。这意味着，在相同预算下，可以承载五到10倍的流量。同时，规模化扩张的单位经济效益也会显著改善，从而避免业务规模越大，企业反而越接近破产。

So, we see kind of those as as two compelling story behind the reason and timing.

因此，我们认为，这就是背后两个极具说服力的理由，也解释了为什么应当在现在采取行动。

Thank you, Lin.

谢谢你，Lynn。

Thanks a lot.

非常感谢。

[applause]

[掌声]