blocks|key|1606596|text|T5模型是否也对BoolQ进行了培训，这会使测试/评估变得困难和可疑，因为后面的测试数据并不是模型中真正不可见的数据。你可以在曲面模型卡和谷歌的原纸中看到它。|type|unstyled|depth|inlineStyleRanges|entityRanges|offset|length|data|1606597|我发现奇怪的是，给预先训练的T5基来自数据集不会产生预期的答案或答案格式。的问题。t5+for+BoolQ有一个精心调整的版本，它提供了一个更可接受的答案。即使使用本文中的精确示例和格式，以SQuAD格式回答问题的预训练模型也存在同样的问题。|1606598|这让我认为，问题回答的微调与其他一些任务不同，这些任务实际上没有包含在模型的发布版本中，或者至少没有足够的效果让模型记住任务是如何工作的。在这种情况下，对其进行微调(再一次/更多)将是有意义的。|1606599|entityMap|0|LINK|mutability|MUTABLE|url|https://huggingface.co/t5-base|1|https://arxiv.org/abs/1905.10044|2|https://huggingface.co/t5-base?text=question%253A%2520is%2520there%2520a%2520now%2520you%2520see%2520me%25203%2520coming%2520out%250A%2509%2509context%253A%2520Now%2520You%2520See%2520Me%2520is%2520a%2520series%2520of%2520heist%2520thriller%2520film%2520written%2520by%2520%250AEd%2520Solomon%252C%2520Boaz%2520Yakin%252C%2520and%2520Edward%2520Ricourt.%2520They%2520focus%2520on%2520the%2520actions%2520of%250A%2520a%2520team%2520of%2520illusionists%2520named%2520%2522The%2520Four%2520Horsemen%2522%2520who%2520pull%2520off%2520near%2520%250Aimpossible%2520heists.%2520The%2520series%2520features%2520an%2520ensemble%2520cast%2520including%2520Jesse%2520%250AEisenberg%252C%2520Mark%2520Ruffalo%252C%2520Woody%2520Harrelson%252C%2520Isla%2520Fisher%252C%2520Dave%2520Franco%252C%2520%250AMichael%2520Caine%252C%2520Lizzy%2520Caplan%252C%2520and%2520Morgan%2520Freeman.%2520The%2520first%2520film%2520was%2520%250Areleased%2520in%25202013%252C%2520while%2520the%2520second%2520was%2520released%2520in%25202016%252C%2520and%2520a%2520third%2520%250Afilm%2520is%2520currently%2520in%2520development%2520and%2520set%2520to%2520be%2520released%2520in%25202019.%2520The%2520%250Aseries%2520has%2520received%2520mixed%2520reviews%2520from%2520critics%2520and%2520audiences%2520and%2520grossed%250A%2520nearly%2520%2524700%2520million%2520worldwide.|3|https://huggingface.co/mrm8488/t5-small-finetuned-boolq?text=question%253A%2520is%2520there%2520a%2520now%2520you%2520see%2520me%25203%2520coming%2520out%250A%2509%2509context%253A%2520Now%2520You%2520See%2520Me%2520is%2520a%2520series%2520of%2520heist%2520thriller%2520film%2520written%2520by%2520%250AEd%2520Solomon%252C%2520Boaz%2520Yakin%252C%2520and%2520Edward%2520Ricourt.%2520They%2520focus%2520on%2520the%2520actions%2520of%250A%2520a%2520team%2520of%2520illusionists%2520named%2520%2522The%2520Four%2520Horsemen%2522%2520who%2520pull%2520off%2520near%2520%250Aimpossible%2520heists.%2520The%2520series%2520features%2520an%2520ensemble%2520cast%2520including%2520Jesse%2520%250AEisenberg%252C%2520Mark%2520Ruffalo%252C%2520Woody%2520Harrelson%252C%2520Isla%2520Fisher%252C%2520Dave%2520Franco%252C%2520%250AMichael%2520Caine%252C%2520Lizzy%2520Caplan%252C%2520and%2520Morgan%2520Freeman.%2520The%2520first%2520film%2520was%2520%250Areleased%2520in%25202013%252C%2520while%2520the%2520second%2520was%2520released%2520in%25202016%252C%2520and%2520a%2520third%2520%250Afilm%2520is%2520currently%2520in%2520development%2520and%2520set%2520to%2520be%2520released%2520in%25202019.%2520The%2520%250Aseries%2520has%2520received%2520mixed%2520reviews%2520from%2520critics%2520and%2520audiences%2520and%2520grossed%250A%2520nearly%2520%2524700%2520million%2520worldwide.|4|https://huggingface.co/t5-base?text=question%253A%2520What%2520does%2520increased%2520oxygen%2520concentrations%2520in%2520the%2520patient%25E2%2580%2599s%2520lungs%2520displace%253F%2520context%253A%2520Hyperbaric%2520%2528high-pressure%2529%2520medicine%2520uses%2520special%2520oxygen%2520chambers%2520to%2520increase%2520the%2520partial%2520pressure%2520of%2520O%25202%2520around%2520the%2520patient%2520and%252C%2520when%2520needed%252C%2520the%2520medical%2520staff.%2520Carbon%2520monoxide%2520poisoning%252C%2520gas%2520gangrene%252C%2520and%2520decompression%2520sickness%2520%2528the%2520%25E2%2580%2599bends%25E2%2580%2599%2529%2520are%2520sometimes%2520treated%2520using%2520these%2520devices.%2520Increased%2520O%25202%2520concentration%2520in%2520the%2520lungs%2520helps%2520to%2520displace%2520carbon%2520monoxide%2520from%2520the%2520heme%2520group%2520of%2520hemoglobin.%2520Oxygen%2520gas%2520is%2520poisonous%2520to%2520the%2520anaerobic%2520bacteria%2520that%2520cause%2520gas%2520gangrene%252C%2520so%2520increasing%2520its%2520partial%2520pressure%2520helps%2520kill%2520them.%2520Decompression%2520sickness%2520occurs%2520in%2520divers%2520who%2520decompress%2520too%2520quickly%2520after%2520a%2520dive%252C%2520resulting%2520in%2520bubbles%2520of%2520inert%2520gas%252C%2520mostly%2520nitrogen%2520and%2520helium%252C%2520forming%2520in%2520their%2520blood.%2520Increasing%2520the%2520pressure%2520of%2520O%25202%2520as%2520soon%2520as%2520possible%2520is%2520part%2520of%2520the%2520treatment.^0|1R|5|0|20|2|1|0|M|F|2|1W|9|3|2M|C|4|0|0^^$0|@$1|2|3|4|5|6|7|X|8|@]|9|@$A|Y|B|Z|1|10]|$A|11|B|12|1|13]]|C|$]]|$1|D|3|E|5|6|7|14|8|@]|9|@$A|15|B|16|1|17]|$A|18|B|19|1|1A]|$A|1B|B|1C|1|1D]]|C|$]]|$1|F|3|G|5|6|7|1E|8|@]|9|@]|C|$]]|$1|H|3|-4|5|6|7|1F|8|@]|9|@]|C|$]]]|I|$J|$5|K|L|M|C|$N|O]]|P|$5|K|L|M|C|$N|Q]]|R|$5|K|L|M|C|$N|S]]|T|$5|K|L|M|C|$N|U]]|V|$5|K|L|M|C|$N|W]]]]

Wasn't the T5 model also trained on BoolQ which would make this difficult and kind of fishy to test/evaluate because the later test data would not really be unseen data for the model? You can see it listed in the <a href="https://huggingface.co/t5-base" rel="nofollow noreferrer">model card on huggingface</a> as well as Google's <a href="https://arxiv.org/abs/1905.10044" rel="nofollow noreferrer">original paper</a>.
What I do find strange is that giving the pretrained T5-base a question from the dataset <a href="https://huggingface.co/t5-base?text=question%3A%20is%20there%20a%20now%20you%20see%20me%203%20coming%20out%0A%09%09context%3A%20Now%20You%20See%20Me%20is%20a%20series%20of%20heist%20thriller%20film%20written%20by%20%0AEd%20Solomon%2C%20Boaz%20Yakin%2C%20and%20Edward%20Ricourt.%20They%20focus%20on%20the%20actions%20of%0A%20a%20team%20of%20illusionists%20named%20%22The%20Four%20Horsemen%22%20who%20pull%20off%20near%20%0Aimpossible%20heists.%20The%20series%20features%20an%20ensemble%20cast%20including%20Jesse%20%0AEisenberg%2C%20Mark%20Ruffalo%2C%20Woody%20Harrelson%2C%20Isla%20Fisher%2C%20Dave%20Franco%2C%20%0AMichael%20Caine%2C%20Lizzy%20Caplan%2C%20and%20Morgan%20Freeman.%20The%20first%20film%20was%20%0Areleased%20in%202013%2C%20while%20the%20second%20was%20released%20in%202016%2C%20and%20a%20third%20%0Afilm%20is%20currently%20in%20development%20and%20set%20to%20be%20released%20in%202019.%20The%20%0Aseries%20has%20received%20mixed%20reviews%20from%20critics%20and%20audiences%20and%20grossed%0A%20nearly%20%24700%20million%20worldwide." rel="nofollow noreferrer">does not yield the expected answer or answer format</a>. There is a fine-tuned version of t5 for BoolQ which gives <a href="https://huggingface.co/mrm8488/t5-small-finetuned-boolq?text=question%3A%20is%20there%20a%20now%20you%20see%20me%203%20coming%20out%0A%09%09context%3A%20Now%20You%20See%20Me%20is%20a%20series%20of%20heist%20thriller%20film%20written%20by%20%0AEd%20Solomon%2C%20Boaz%20Yakin%2C%20and%20Edward%20Ricourt.%20They%20focus%20on%20the%20actions%20of%0A%20a%20team%20of%20illusionists%20named%20%22The%20Four%20Horsemen%22%20who%20pull%20off%20near%20%0Aimpossible%20heists.%20The%20series%20features%20an%20ensemble%20cast%20including%20Jesse%20%0AEisenberg%2C%20Mark%20Ruffalo%2C%20Woody%20Harrelson%2C%20Isla%20Fisher%2C%20Dave%20Franco%2C%20%0AMichael%20Caine%2C%20Lizzy%20Caplan%2C%20and%20Morgan%20Freeman.%20The%20first%20film%20was%20%0Areleased%20in%202013%2C%20while%20the%20second%20was%20released%20in%202016%2C%20and%20a%20third%20%0Afilm%20is%20currently%20in%20development%20and%20set%20to%20be%20released%20in%202019.%20The%20%0Aseries%20has%20received%20mixed%20reviews%20from%20critics%20and%20audiences%20and%20grossed%0A%20nearly%20%24700%20million%20worldwide." rel="nofollow noreferrer">a more acceptable answer</a>. Same problem with the pretrained model for <a href="https://huggingface.co/t5-base?text=question%3A%20What%20does%20increased%20oxygen%20concentrations%20in%20the%20patient%E2%80%99s%20lungs%20displace%3F%20context%3A%20Hyperbaric%20%28high-pressure%29%20medicine%20uses%20special%20oxygen%20chambers%20to%20increase%20the%20partial%20pressure%20of%20O%202%20around%20the%20patient%20and%2C%20when%20needed%2C%20the%20medical%20staff.%20Carbon%20monoxide%20poisoning%2C%20gas%20gangrene%2C%20and%20decompression%20sickness%20%28the%20%E2%80%99bends%E2%80%99%29%20are%20sometimes%20treated%20using%20these%20devices.%20Increased%20O%202%20concentration%20in%20the%20lungs%20helps%20to%20displace%20carbon%20monoxide%20from%20the%20heme%20group%20of%20hemoglobin.%20Oxygen%20gas%20is%20poisonous%20to%20the%20anaerobic%20bacteria%20that%20cause%20gas%20gangrene%2C%20so%20increasing%20its%20partial%20pressure%20helps%20kill%20them.%20Decompression%20sickness%20occurs%20in%20divers%20who%20decompress%20too%20quickly%20after%20a%20dive%2C%20resulting%20in%20bubbles%20of%20inert%20gas%2C%20mostly%20nitrogen%20and%20helium%2C%20forming%20in%20their%20blood.%20Increasing%20the%20pressure%20of%20O%202%20as%20soon%20as%20possible%20is%20part%20of%20the%20treatment." rel="nofollow noreferrer">Question answering in the SQuAD format</a> even when using the exact example and format from the paper.
Which leads me to think the fine-tuning on question answering is unlike some other tasks not actually included in the released version of the model or at least does not seem to have enough of an effect for the model to remember how the task works. In which case fine-tuning on it (again/more) would make sense again.

I want to use the pre-trained T5 model <a href="https://huggingface.co/docs/transformers/model_doc/t5" rel="nofollow noreferrer">https://huggingface.co/docs/transformers/model_doc/t5</a> on the task of Question Answering on the <a href="https://huggingface.co/datasets/boolq" rel="nofollow noreferrer">https://huggingface.co/datasets/boolq</a> knowing that my inputs will be the passage and the question and the output is the boolean true or false that is the answer for the question.
I have seen some people tuning the model to this specific task. But, I want to know if there is a way to do it with pre-trained model to get some outputs and then compare them with the model after tuning.
Thanks!

Question Answering with pre-trained model T5

翻译质量差，导致语言生硬或混乱。

没有提供实际的解决方法或示例。

解答不清晰，无法理解或解决问题。

页面排版不美观，阅读体验差。

文章

问答

视频

学习中心

腾讯云实验室

直播

竞赛

腾讯云代码分析专区

腾讯iOA零信任安全管理系统专区

腾讯云架构师技术同盟交流圈

腾讯云数据库专区

腾讯云顾问专区

腾讯云原生专区

腾讯混元专区

腾讯云TCE专区

腾讯云Lighthouse专区

腾讯云HAI专区

腾讯云Edgeone专区

腾讯云存储专区

腾讯云智能专区

腾讯轻联专区 

腾讯云开发专区

TAPD专区

腾讯轻量云游戏服专区

腾讯云最具价值专家

腾讯云架构师技术同盟

腾讯云创作之星

腾讯云开发者先锋

腾讯云代码助手

云原生构建

TAPD 敏捷项目管理

Cloud Studio

SDK中心

API中心

命令行工具

涵盖代码开发、场景应用、自动测试全流程，助你从零构建专属AI助手

一站式MCP教程库，解锁AI应用新玩法

我想使用预先训练过的T5模型来执行在上回答问题的任务，知道我的输入是段落和问题，输出是布尔值，真假，这是问题的答案。我见过一些人把模型调到这个特定的任务。但是，我想知道是否有办法用预先训练过的模型来获得一些输出，然后在调优后将它们与模型进行比较。谢谢!

用预训练模型T5回答问题-腾讯云开发者社区-腾讯云

问用预训练模型T5回答问题
EN

回答 1

Stack Overflow用户

社区

活动

圈层

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问用预训练模型T5回答问题EN