Luyện nói tiếng Anh bằng Shadowing qua video: Every Size Local AI In 24 Minutes

Chọn chế độ luyện tập

Mỗi chế độ mở một phòng luyện tập toàn màn hình.

401 câu

Nếu các câu quá ngắn hoặc quá dài, hãy bấm Edit để chỉnh sửa.

Hello!

Today we're going to be running every size AI locally with the hardware

that we have here and see what we can do with it.

Hopefully give you guys some ideas for cool things that you can build with local AI too.

Let's go!

A portion of this video is sponsored by Crusoe.

So I'm going to be grouping the hardware that I have here into different classes, different memory classes.

So the first class are chips.

So chips, as they are called, are literally just chips, just computer chips.

They have no operating system, no file system, and no separate memory stick either.

Literally everything is just built into these boards.

So we try to run an AR model on these.

We're not even installing it.

We just literally stick it on here, and if it fits, it sits, and we can run it. Or not.

And in my case, the smallest one that I have here is an Arduino R4 Uno.

It has only 32 kilobytes of RAM.

I will put the rest of the stats on screen, which is big enough to fit.

Drumroll please.

No models.

That's right, nothing fits on here.

Alas.

I actually just got this Arduino Uno R4 as part of a set

because I'm trying to learn hardware and it unfortunately does not fit any AI models.

But do not be fooled, that does not mean I cannot use this with local AI.

I actually can since it has a Wi -Fi functionality so I can connect with some of my bigger devices.

But first let's actually talk about something that can actually fit an AI model.

Isn't this shocking.

It is so tiny.

This is the ESP32 S3 and it has 8 megabytes of RAM.

I will put the rest of the stats on screen

and I got this for around $20 USD although this one does come with a screen

so if you get it without the screen you're looking at like two to five dollars.

So cheap and you can actually run AI models on here.

Two AI models actually.

Both of them are called tiny story models which as its name suggests allows you to run tiny stories.

Let me show you what it looks click.

So this one is the Tiny Stories 1MB model, and then here is the Tiny Stories 3MB model.

You can run both of these very comfortably.

So yes, it can run the Tiny Stories 260K model and the Tiny Stories 3M model very comfortably.

And honestly, this is just the start, getting a model running.

You can make it so much cooler by adding on different modules to it.

For example, you can make it into a little game called Scrambled vs Not Scrambled.

Might be useful if you're learning english grammar

and you can attach led lights

that represent the mood of the story that's being produced

so many cool things that you can do maybe i will make another video showcasing stuff

that you can build using these little devices

but alas we do not have time in this video

so here is a summary slide for the models

that we can run on the esp32 and some ideas for ai projects

that you can build take a screenshot in the free guide

in the description i will also link a little quick start guide for how to get started building

so these tiny stories models are only able to do the

very basics of predicting the next word in order to form a coherent story

which is still very impressive actually given how small it is

and you can build some really cool things on top of

that too but it is not capable of holding a conversation like a chat

and it's not able to run any multi modality models like audio music video

or images for that we're going to need to move on to the next class of devices

which is the mini computers category

this is a raspberry pi the pi 5

and this is my personal iphone 16 now these are proper

computing devices they run an operating system the raspberry pi runs linux

and my iphone runs the ios

and they're able to do things like manage memory file systems

and run multiple programs and

so you can actually have multiple models as long as they fit into these devices

but what is really interesting thing is that both of these actually have 8 gigs of RAM, yet they're not able to run the same kinds of models.

I'm going to give you guys a quick lesson on AR hardware in just a little bit, but first I want to show you what these guys can do.

This is the Raspberry Pi, the Pi 5, and even though it looks very small, it is very mighty because it is a full -blown mini computer.

Let me show you what you can do with it.

What color is an octopus?

Octopuses come in a variety of colors, but they are usually camouflage blending with their surroundings they can be blue green red

or even white depending on the species first is using a

speech to text ai model called whisper in order to listen

to it then i have the quen 3 1 .7b model it's a large language model

which is an lm able to interpret it

and then come out with a response and then the piper model

which is a tiny text to speech model is able to actually answer it out loud on these speakers isn't

that crazy i'm able to have full conversations with my Pi by chaining together three different AI models, all on this tiny little device.

But wait, that's not all.

I can also connect a webcam that grabs a frame, takes a picture, and I can have a tiny visual language model called Moondream,

describe it, and the tiny text -to -speech model Piper is able to describe what it sees.

Here is the camera, the webcam attached to the Raspberry Pi.

The image features a woman sitting in a chair looking at her cell phone.

She is wearing a black shirt and is surrounded by a variety of books and other items.

There are at least 13 books scattered around the room, some of which are placed on a bookshelf.

A backpack can be seen near the woman.

and a potted plant is located close to her the scene suggests

that the woman might be engaged in reading

or studying as she is surrounded by numerous books and other items

so yeah you can pretty much run micro to small large

language models very easily on the raspberry pi as well as

the tiny tier visual language models tiny text to speech models

and small but decent speech to text models it does struggle to run mid -sized large language models

but if you really want to you kind of can do it

and it can kind of barely run small code tier models as well.

But it really cannot do image generation, music generation, or video generation.

Not actually because a small image generation model wouldn't fit on here.

100

It does.

101

The problem is that it doesn't have any GPUs so it's just like painfully slow to do that.

102

And I will explain GPUs a little bit later.

103

But it does make up for it because you can build using the Raspberry Pi.

104

You can add so many different types of accessories

105

and sensors to your raspberry pi in fact i believe the

106

hugging face robots are running off a raspberry pi as well

107

so yes i'm going to put on a summary slide now including some ideas of things

108

that you can build with a raspberry pi

109

and please do check out the free guide linked below for

110

a quick start guide now let's talk about the iphone 16

111

which also has eight gigs of ram same as the raspberry pi

112

but it is so different for one

113

because it is running on ios apple does limit each app

114

to only be able to use around four to five gigs of ram so even though technically has 8 gigs of RAM, you actually aren't able to run the mid -sized level of models from 7 to 14 beat

115

that you can't do using the Raspberry Pi because like Apple just doesn't let you do it, unfortunately.

116

Also because of Apple reasons, you can't just be like putting on things that you wish onto your iPhone.

117

You have to do it through Apple's official system in

118

which you need to install an app and the app allows you to download, install, and run your local AI models.

119

To get the Gemma 4 model, you need to download the AI Edge Gallery from Google

120

and then download the models locally on your phone this phone

121

is able to run these small plus visual language models meaning

122

it can analyze images of the 1 to 4b range very

123

comfortably since it's only 1 .5 gigs for the image generation you need to download an app called draw things

124

and you can download

125

and run any small image generation models at around the 1b

126

range with no problem like stable diffusion 1 .5 for example

127

is 2 gigs you can run it very comfortably you can also run medium

128

and large speech text models like the whisper large very comfortably

129

and all text -to -speech models too and finally you can run a model called a detection model.

130

It's actually very cool.

131

People don't really talk about it very much for some reason.

132

You can download an app called Ultralytics and download and run the YOLO 11n model.

133

The YOLO 11n model allows you to do object recognition to understand the scene

134

and it's the underlying technology behind all things from facial recognition on your devices to self driving cars

135

because you know you gotta be able to recognize different objects as we're driving.

136

We don't really talk about this class of models very much, these detection models, but they're actually very very useful.

137

Isn't that really cool?

138

I'm going to put on screen now a summary slide including some ideas of things

139

that you can build using your phone with local AI.

140

A small lesson on AI hardware.

141

Now think about a device, any device.

142

It is like a restaurant kitchen.

143

Downstairs we have a walk -in freezer and a storage space.

144

This is where we store all the ingredients, all of the things.

145

Then we have the prep counter.

146

This is where we load up all the ingredients that we need right now for the dish that we want to make.

147

Now let's talk about people who work in the kitchen we

148

have the head chef the head chef is very experienced they can make any sort of dish

149

but just one head chef there is also a brigade of

150

line cooks these line cooks are less experienced less versatile

151

but we have a lot of them that can help us out

152

and finally we have specialized machines bolted to the walls like say an onion machine

153

that only chops onions or like a noodle slicing machine

154

that just automatically sizes noodles i don't know

155

if you guys have seen those before

156

but basically these machines can only do one thing it does

157

that one thing really well and really fast and it uses almost no electricity Now, between the prep counter and the chefs and the cooks and the machines, there is a conveyor belt passageway.

158

This is how we get the ingredients from the prep counter to the chefs, cooks, and machines to process them into, hopefully, a meal.

159

So say, for example, we want to prepare a salad.

160

Well, first, gotta go get all of the salad stuff from the storage in the freezer and put them onto prep counter, like your carrots, lettuce, olives, chicken, other salad stuff.

161

Then we put it on the conveyor belt and it gets passed through

162

and the head chef starts giving out orders tells all the line cooks to be chopping up all the vegetables

163

and also starts up the onion machine to be chopping up onions

164

while he himself is only say like mixing the sauce

165

because that's like the tricky part and then after a few minutes everything comes together

166

and voila we get a salad yay amazing

167

so this is a great analogy for how ai models work on devices

168

so this time what we want to be cooking is a model

169

so we walk down to our freezer downstairs except we call it storage

170

and we grab our model and we put it onto our prep counter which we call the RAM.

171

Then we pass it along the conveyor belt

172

which we call a memory bus

173

and hand it to our head chef called the CPU who

174

then also directs the line cooks to start cooking up parts of the model.

175

The line cooks are called the GPU and also start up the little specialized bolted machines, the onion machine which we call the NPU.

176

And after all this cooking your model is able to produce something like respond to your question of hi how are you?

177

The model then can say hello I am good how are you? okay makes sense

178

so every time we want to be working with the model

179

we need to be going through this entire process of cooking great now you understand the major components of hardware

180

when it comes to ai models however remember the question

181

that we asked earlier why is it

182

that a raspberry pi has 8 gigs of ram

183

and an iphone has 8 gigs rams too but they are

184

so different models run an iphone are much faster

185

and it's able to do image generation while the raspberry pi cannot well

186

that is because these two devices have the same amount of ram

187

which means that the same amount of capacity But they do not have the same amount of bandwidth,

188

which is referring to the conveyor belt between the prep counter and our cook and our chefs.

189

You see, for most large language models, they are bottlenecked by how quickly you can deliver them from the RAM,

190

the prep counter, to the chefs and the cooks and the machines to process the model and get them to be running.

191

On a Raspberry Pi, this conveyor belt, the memory bus, is quite narrow and slow.

192

So it physically takes longer to be able to be processed.

193

And on top of that, a Raspberry Pi only has a CPU without any GPUs or NPUs.

194

So it's basically like having a single head chef, very capable, but only a single pair of hands having to do everything to run this model.

195

That's why it is really, really slow.

196

While for the iPhone, it has CPUs, GPUs, and NPUs.

197

That's why it's able to process things so much faster.

198

So that explains why it is that things run so slowly on the Pi, but run so quickly on iPhone.

199

But what about image generation?

200

The Pi simply cannot do image generation.

201

And that again is because of the lack of GPUs you see

202

when it comes to image video and music generation it is really processing heavy so

203

if you just only have cpus no matter how hard they

204

work it's just not enough to be able to generate music images

205

and videos so you really got to have some gpus like the iphone to be able to do

206

that we will return to this analogy a little bit later as i explain more devices

207

but for now amazing you now have an understanding of ai

208

hardware yay i'll put on screen now a little pop quiz answer these questions

209

and put them into the comments to make sure

210

that you're paying attention of course what i said about ai

211

hardware is much more simplified than it actually is there's concepts like vram unified memory like

212

so much more which i'm not going to go into too much more detail about in this video

213

because i'm sure this video is already very long

214

but this is like the general introduction

215

that will help you a lot in figuring out what kind of models can run on what kind of devices

216

so as you guys can see from this video i have been very into tinkering

217

and running open source models And the next obvious step is also tinkering and customizing the open source models themselves.

218

For that, Crusoe Intelligence Foundry has been amazing.

219

It primarily gives you two things.

220

Serverless inference to run leading open models without touching any of the underlying GPU infra.

221

And serverless fine -tuning to customize the open models on your own data without having to own or manage any hardware.

222

You can fine -tune a model, deploy it in one click, and start calling it with an API in minutes.

223

Seriously, the first time that I did this, I almost shed a tear.

224

There is no clusters to provision or crazy long setups.

225

I cannot even explain how painful this used to be.

226

So for today, anybody that's building with open models who don't want the infrastructure overhead to eat into actual build time, this is a really clean and easy path to get started.

227

And now when you sign up, you will get $5 of free credit.

228

You can now try out Caruso, link is in the description.

229

Thank you so much, Caruso, for sponsoring this portion of the video.

230

Now, back to the video.

231

Okay, now let's move on to the next category of devices, which is personal computers, such as my MacBook Pro that I'm more using right now.

232

So my MacBook Pro is 32 gigs, but of course there are other machines in this category as well, like my older machine, which is a MacBook Air,

233

and of course, Windows laptops and Linux laptops too.

234

They all belong in these portable, personal computer, personal machines category.

235

So the reason why I got a MacBook Pro with 32 gigs of RAM is

236

because this is the machine that pretty much allows me to run any category of AI models.

237

I can run large language models, coding models, visual language models for visual analysis, image generation, video generation, text -to -speech,

238

speech -to -text, audio generation, voice cloning, and music generation.

239

All of this I can run locally.

240

And not only can I run each of these models locally by itself, I'm also able to use these models in different ways by using it with different types of software.

241

For example, with my Hermes agent, which I use to run many aspects of my life and my work, I can run pretty much any mid -size to large -size local model, like the Quen 3 14B,

242

16B, 32B, some of my go -tos in the Quen family.

243

And I can run the Quen 3 Coder 30B, locally using also an open -source coding harness like OpenCode, and it works pretty good.

244

My favorite for image generation is FluxDev.

245

I'm going to put a summary slide now of all the different types of models

246

that you can run in this personal computer, personal machine category, as well as some ideas of things that you can build with these models.

247

Now, I know what you're thinking already, but Tina, I don't have a 32 gig MacBook Pro.

248

Understandable.

249

It is really expensive.

250

I only got this machine because it is literally my job to be testing out models and things, right?

251

Don't worry, I got you.

252

I'm also gonna put on screen now the models that you can run if you have.

253

8 gigs of ram and 16 gigs of ram these are the general ram sizes of laptops

254

that most people have but if you do have something that's not like 8 gigs

255

or 16 gigs literally you have something smaller bigger something in the middle

256

that is okay as well

257

because i'm also teach you a formula for calculating roughly the size of model

258

that you can run on your machine giving the amount of ram

259

that you have okay ready the equation is

260

that take the total amount of ram that your machine says it has subtract

261

that by 25 of it because this is used up by like other stuff

262

and the rest is the actual usable amount of RAM that you have.

263

Divided by 0 .6 and that equals the amount of parameters that your model can have.

264

For example, my MacBook Pro says it has 32 gigs of RAM.

265

So I subtract 25 % of that, which means I have 24 gigs of usable RAM.

266

Now I do 24 divided by 0 .6 to get 40.

267

So my MacBook Pro can fit a model of up to 40 billion parameters.

268

That is a very rough way of figuring it out.

269

Of course, you can also just ask one of your favorite AI chatbots.

270

Hey, this is my computer

271

and what kind of models can i fit on it you know

272

that works too but in any case i hope

273

that is helpful i want to now move on to the next category of devices home servers the two

274

that i have to demo right now is the mac studio

275

and the amd ryzen ai halo all right the mac studio

276

and the amd ryzen these are examples of the home server class there are a lot of other machines

277

that fit into this category i would say the cheapest entry

278

level here would be a mac mini although those are really hard to get these days.

279

I actually only got the Mac Studio because I couldn't get a Mac Mini.

280

Anyways, at this tier, what is the most attractive and why you would get something like this is for two reasons.

281

The first one is that it is a dedicated always -on machine.

282

So unlike your laptop, which you gotta carry around and close in, stuff like that, and when you close it, your, you know, your models start working in case you were doing something with it.

283

But the home servers, these machines are always on.

284

They are designed to be always on, so they don't close down.

285

You can constantly be doing stuff and running your AM models 24 -7.

286

That's the first attractive part.

287

And the second reason is that generally these home server devices do have more capacity

288

and bandwidth so it can run a bigger models

289

and do it faster let's first talk about the mac studio

290

here it has 64 gigs of ram pop quiz it says

291

it has 64 gigs of ram how much of it is actually usable

292

and what is the largest model in billions of parameters

293

that can fit on my mac studio very paying attention put it into the comments amazing hope you got

294

that my mac studio is where i like to run my local models to use for my local ai agents

295

and also allow other devices remember this device the arduino uno r4

296

that couldn't run anything by itself i also like to be able to connect this with my mac studio

297

so it can use the models

298

that are being run on my mac studio this arduino uno

299

is using the quen 3 .6 35b model on this mac

300

studio in order to display encouraging happy messages write into the comments what this says if you can read it.

301

Yay!

302

That's really cool, right?

303

Really with this Mac Studio and all devices that are bigger and more powerful than this, it's not about unlocking new categories of AI models at this point.

304

It's just about being able to run bigger models, higher quality things to get higher quality responses, and faster.

305

In the case of the Mac Studio, I can run large tier, large language models between 27 to 32 billion parameters, as well as large code models,

306

large image models, and large music generation models.

307

for video generation i can now run mid -tier video generation like the one 2 .2

308

but it is slow it is really slow still i'm going

309

to put a summary slide now with all the models

310

that you can run with a mac studio as well as some ideas for things you can build

311

and let's move on to the amd halo

312

which is 128 gigabytes of advertised ram now with 128 gigabytes of advertised ram Around 96 gigabytes of usable RAM,

313

it is now able to comfortably run extra large models in the 70B range

314

and Frontier models in the 100 to 250B range.

315

We can see that Llama 3 .3 70B is able to run fine

316

and GPT -OSS 120B is also able to run fine.

317

It's also able to run extra large tier of coding models like the DeepSeat Coder V2.

318

Great.

319

So we can see that it can run bigger models now.

320

Makes sense.

321

However, what I find the most useful of having something like this

322

is the ability of having a lot of different models loaded simultaneously

323

and working simultaneously because it has that massive amount of memory, right?

324

So I could be using a coding agent, the large language models, and doing like image video generation all at the same time.

325

I'm going to put on screen now all the models

326

that you can run with a device like the AMD Halo and ideas of what you can build.

327

But I do want to make a caveat here though.

328

This machine, even though it has very large capacity, it actually has pretty small bandwidth, which means that it is slow,

329

especially when it comes to image, video, and music generation.

330

You can do it.

331

It's just going to do it really, really slow, which is why I want to introduce you to the final category that we're going to cover today, which is GPUs.

332

First, let me explain GPUs a little bit more remember the analogy

333

that we had earlier the restaurant kitchen analogy we said

334

that gpus is like having a brigade of line cooks

335

because we have so many of them they're able to work really really quickly

336

and it do compute very fast now this matters the most

337

when it comes to music generation image generation and video generation

338

because to be able to produce these you just need like a lot of processing power

339

so you need a lot of line cooks in the kitchen the bad news is

340

that all the devices that i went through earlier they do of course have gpus

341

which is why they can do things like music generation image generation

342

and video generation but they do it really really slowly

343

because they're not specialized at doing this but the good news is

344

that there is a way to be really good at processing

345

and be able to do really fast and really good image music

346

and video generation and

347

that is by getting more discrete gpus you see a gpu actually isn't just line cooks it's actually a bundle

348

when you get another gpu you get cooks

349

but they also come with their own private prep counters

350

and a very wide conveyor belt the memory pass so

351

when you get more gpus you get more cooks who have their own private counter

352

and their very own very wide conveyor belt the memory pass

353

so they're able to get the ingredients really fast it's basically like a boost for the line cooks

354

so you're able to process things really really quickly cook your models really quickly i hope

355

that analogy makes sense it's getting really late

356

so there are a few ways to get more gpus unfortunately for all the devices

357

that we covered earlier you cannot add more gpus to them just like they're not built

358

that way but if you had a gaming pc for example that would have a GPU.

359

I do not have a gaming PC and I'm not about to go buy one right now because those are expensive.

360

But if you do have one, good news for you.

361

You have a GPU and you can actually add more GPUs as well.

362

So I'm going to cheat a little bit, okay?

363

Because I do want to show you what it's like to run these multi -modality generation models on these specialized GPUs.

364

So I'm actually going to rent an RTX 4090.

365

This is a pretty high -end GPU that you might have in a high -end gaming set.

366

Let's do this quickly because it's costing me by the hour.

367

We're going to log in here and then I'm going to run an image generation model.

368

You can see that it is really, really quickly.

369

And here is a video generation as well.

370

Really, really quick.

371

Now, the downside of this GPU is that it only has 24 gigs of memory.

372

So you can think about it almost like the opposite of the AMD device

373

because the AMD device has a lot of memory, has very big capacity, but low bandwidth while the RTX 490 has smaller capacity, only 24 gigs, but much larger bandwidth.

374

So in the end, you end up being able to generate a lot of these things and run these models, but you can't really fit very large size models on it.

375

All right, I'm going to now put a summary slide for the models

376

that you can run on this device and the things that you can build with it.

377

And that finally leads me to the final boss, the 8X H100 GPU.

378

Now this is top tier, 640 gigs of RAM, massive capacity and massive bandwidth.

379

Now I really gotta move fast on this one

380

because it's costing me so much money every minute but for you guys it's worth it.

381

On here not only can we run ultra large tier large

382

language models up to 700 billion parameter plus ultra large coding

383

models 480 billion parameters plus visual language processing as well and all types of multimodal generation.

384

Specifically let's check out video generation.

385

Using the minimax h3 model we can see that it's able to generate minutes of video and it does it so quickly.

386

This is pretty much what you can run assuming

387

that you're not like a literal model development company or just have like a lot of money.

388

But it really is comparable at this point to the frontier models that you're paying for through cloud APIs.

389

You see everything up to this GPU is kind of just forcing you to choose between like capacity

390

and bandwidth right like the AMD Halo is all about big capacity below bandwidth

391

and the RTX 490 is all about small capacity and large bandwidth

392

but this is when you get both you get very big capacity

393

and very big bandwidth

394

and now for the 8H100 rented I'm gonna put a summary slide now of the models

395

that you can run with it and the things

396

that you can build this is like aspirational class great amazing wow thank you

397

so much for watching until the end of this video i hope this was insightful interesting helpful

398

and it has inspired you to want to build things with

399

local ai as well i'm going to put on screen now

400

a little quiz please answer these questions in the comments below to help you retain all of the information

401

that we have covered today thank you so much and I will see you guys in the next video or live stream.

Tải Ứng Dụng

Mọi thứ bạn cần để nói trôi chảy

Chấm điểm AIChấm từng câu nói

Luyện IPAThành thạo mọi âm

Từ vựngMở rộng vốn từ

Game Từ vựngHọc qua chơi

Quét QR

Download on theApp Store Get it onGoogle Play

Cùng kênh

FOR YOU

Video gợi ý

63 câu

5 open source tools that replaced my $320/mo AI stack...

Business EnglishUniversity StudentsWorking Adults

41 câu

Making a 360° Controller Smaller than a Penny!

Daily ConversationUniversity StudentsWorking Adults

190 câu

Supervised Learning: Crash Course AI #2

Business EnglishUniversity StudentsWorking Adults

192 câu

Making RAM at Home

Business EnglishUniversity StudentsWorking Adults

12 câu

Rapid Engineering With AI

Business EnglishUniversity StudentsWorking Adults

102 câu

World Models explained in 10min..

Business EnglishUniversity StudentsWorking Adults

144 câu

WWDC26: Run local agentic AI on the Mac using MLX | Apple

Business EnglishUniversity StudentsWorking Adults

394 câu

Chip War 2.0: The Global Battle for Semiconductor Supremacy│Chris Miller (The Author of “Chip War")

IELTS PrepBusiness EnglishUniversity Students

43 câu

P1 机器学习的应用【2024公认最好的 | 吴恩达机器学习 | 教程 | Machine Learning Specialization(超爽中英!)】

Business EnglishUniversity StudentsWorking Adults

33 câu

PyTorch in 100 Seconds

Business EnglishUniversity StudentsWorking Adults

179 câu

Understand AI in 14 minutes – with Anthropic's Chloe Lubinski [ARC 2026]

Business EnglishDaily ConversationUniversity Students

738 câu

The Closest Thing We Have to Alien Technology

Business EnglishUniversity StudentsWorking Adults

Từ vựng và ghi chú luyện nói cho bài học này

Bài luyện nói trình độ C1 này dựa trên video “Every Size Local AI In 24 Minutes”. Những từ được nhắc lại nhiều nhất trong bài: model, generation, gpu, ram, device. Video này có 401 câu và 5239 từ để luyện shadowing. Phần lời nói dài 24:22. Người nói nói nhanh, khoảng 215 từ mỗi phút, nên sẽ có nhiều chỗ nối âm và nuốt âm. 83% số từ nằm trong 3.000 từ tiếng Anh thông dụng nhất; phần còn lại nên xem trước khi luyện.

Từ vựng quan trọng trong video

15 từ khó nhất trong video, kèm phiên âm và nghĩa:

Từ	Phiên âm	Nghĩa
gig danh từ	/ɡɪɡ/	biểu diễn, nhạc sống
preposition danh từ	/ˌpɹɛpəˈzɪʃən/	giới từ
analogy danh từ	/əˈnæləd͡ʒi/	loại suy
onion danh từ	/ˈʌn.jən/	hành tây, hành
usable tính từ	/ˈjuː.zə.bəl/	dùng được, khả dụng
ingredient danh từ	/ɪnˈɡɹiːdi.ənt/	thành phần
freezer danh từ	/ˈfɹizɚ/	máy ướp lạnh, máy làm kem
quiz danh từ	/kwɪz/	đố vui
subtract động từ	/səbˈtɹækt/	trừ
webcam danh từ	/ˈwɛbˌkæm/	webcam
chop động từ	/t͡ʃɑp/	chặt, chém
surround động từ	/səˈɹaʊnd/	bao vây
bolt danh từ	/boʊlt/	bu lông
bookshelf danh từ	/ˈbʊkˌʃɛlf/	giá sách, kệ sách
noodle danh từ	/ˈnu.dəl/	bánh, bún

Cụm động từ bạn sẽ nghe

Từ	Phiên âm	Nghĩa
check out động từ		tính tiền, trả phòng
close down động từ	/kləʊzˈdaʊn/	đóng cửa

Những câu đáng luyện lại

Các câu ngắn, trọn ý trong video mà bạn có thể dùng lại khi giao tiếp hằng ngày:

We're not even installing it.
Don't worry, I got you.
Specifically let's check out video generation.

Ngữ pháp trong video

Những cấu trúc người nói dùng nhiều nhất, kèm đúng cụm từ trong video:

Cấu trúc	Trong video
Câu bị động be + quá khứ phân từ — nhấn vào việc xảy ra, không phải người làm	is sponsored · are called · is just built
Mệnh đề quan hệ who / which + mệnh đề — thêm thông tin về người hoặc vật	screen, which is · people who work · devices, which is
Thì hiện tại hoàn thành have/has + quá khứ phân từ — việc đã xảy ra nhưng còn liên quan đến hiện tại	have specialized · have seen · have been

Phát âm cần chú ý

Người nói dùng 69 dạng rút gọn, ví dụ I'm, don't, they're. Hãy nói theo dạng ngắn đúng như bạn nghe.

Âm “sh” và “zh”: preposition /ˌpɹɛpəˈzɪʃən/, specialize /ˈspɛʃəˌlaɪz/, detection /dɪˈtɛk.ʃən/, aspirational /ˌæspəˈɹeɪʃənəl/, bookshelf /ˈbʊkˌʃɛlf/
Từ dài — đặt trọng âm cho đúng: preposition /ˌpɹɛpəˈzɪʃən/, parameter /pəˈɹæm.ə.tɚ/, analogy /əˈnæləd͡ʒi/, ingredient /ɪnˈɡɹiːdi.ənt/, aspirational /ˌæspəˈɹeɪʃənəl/

Những âm người Việt thường gặp khó:

Cụm phụ âm cuối — đọc đủ từng âm, đừng nuốt âm cuối: bandwidth /ˈbændwɪdθ/, ingredient /ɪnˈɡɹiːdi.ənt/, subtract /səbˈtɹækt/, downstairs /daʊnˈstɛɹz/, surround /səˈɹaʊnd/
Âm /tʃ/, /dʒ/ và /ʒ/ — tiếng Việt không có, đừng đọc thành “ch” hay “gi”: analogy /əˈnæləd͡ʒi/, chop /t͡ʃɑp/, camouflage /ˈkæ.məˌflɑːʒ/, chatbot /ˈtʃætbɑt/, diffusion /dɪˈfju.ʒən/
Âm /æ/ — mở miệng rộng hơn âm “e”: bandwidth /ˈbændwɪdθ/, parameter /pəˈɹæm.ə.tɚ/, analogy /əˈnæləd͡ʒi/, subtract /səbˈtɹækt/, webcam /ˈwɛbˌkæm/

Cách luyện với video này

Nghe hết video một lần, chưa cần nói, và ghi lại những từ bạn chưa biết.
Bắt đầu ở tốc độ 0,75×, nói đuổi từng câu, rồi quay lại tốc độ bình thường khi đã quen.
Ghi âm giọng mình rồi so với bản gốc, chú ý các từ như gig, preposition, analogy.

Phương Pháp Shadowing Là Gì?

Shadowing là kỹ thuật học ngôn ngữ có cơ sở khoa học, ban đầu được phát triển cho chương trình đào tạo phiên dịch viên chuyên nghiệp và được phổ biến rộng rãi bởi nhà đa ngôn ngữ học Dr. Alexander Arguelles. Nguyên lý cốt lõi đơn giản nhưng cực kỳ hiệu quả: bạn nghe tiếng Anh của người bản xứ và lặp lại to ngay lập tức — như một "cái bóng" (shadow) đuổi theo người nói với độ trễ chỉ 1–2 giây. Khác với luyện ngữ pháp hay học từ vựng bị động, Shadowing buộc não bộ và cơ miệng phải đồng thời xử lý và tái tạo ngôn ngữ thực tế. Các nghiên cứu khoa học xác nhận phương pháp này cải thiện đáng kể phát âm, ngữ điệu, nhịp điệu, nối âm, kỹ năng nghe và độ lưu loát khi nói — đặc biệt hiệu quả cho người luyện IELTS Speaking và muốn giao tiếp tiếng Anh tự nhiên như người bản ngữ.

Phương pháp shadowing: đọc hướng dẫn từng bước đầy đủ →

Luyện nói tiếng Anh bằng Shadowing qua video: Every Size Local AI In 24 Minutes

Chọn chế độ luyện tập

Tải Ứng Dụng

How To Learn To Code In 2026

Learn 10X Faster In 10 Minutes

Updated Essential AI Skills For 2026

My Full Claude Cowork Setup (steal my workflows!)

Essential AI Skills For 2026

Here's the AI Roadmap You Actually Need

Best Upcoming Jobs (2026 - 2030)

5 open source tools that replaced my $320/mo AI stack...

Making a 360° Controller Smaller than a Penny!

Supervised Learning: Crash Course AI #2

Making RAM at Home

Rapid Engineering With AI

World Models explained in 10min..

WWDC26: Run local agentic AI on the Mac using MLX | Apple

Chip War 2.0: The Global Battle for Semiconductor Supremacy│Chris Miller (The Author of “Chip War")

P1 机器学习的应用【2024公认最好的 | 吴恩达机器学习 | 教程 | Machine Learning Specialization(超爽中英!)】

PyTorch in 100 Seconds

Understand AI in 14 minutes – with Anthropic's Chloe Lubinski [ARC 2026]

The Closest Thing We Have to Alien Technology

Từ vựng và ghi chú luyện nói cho bài học này

Từ vựng quan trọng trong video

Cụm động từ bạn sẽ nghe

Những câu đáng luyện lại

Ngữ pháp trong video

Phát âm cần chú ý

Cách luyện với video này

Phương Pháp Shadowing Là Gì?