Tag Archives: generation’

Kia Nurse hopes Raptors’ all-women broadcast could mirror WNBA, inspire next generation

If they see it, they can be it.

That’s the notion Canadian WNBA player and TSN commentator Kia Nurse hopes will inspire young women as the NBA’s first all-women broadcast team prepares to call the Raptors’ game against the Denver Nuggets Wednesday. 

Nurse hopes the influence of the first-of-its kind broadcast is similar to what she’s experienced recently as a player for the Women’s National Basketball Association, a league that’s had a huge impact with its social activism.

Some examples include their efforts to help the Democrats win a Senate seat in Georgia and the league’s #SayHerName campaign that created awareness about the police shooting of Breonna Taylor. Maya Moore, one the WNBA’s most famous players, took a sabbatical from her basketball career to help free a wrongfully convicted man who is now her husband.

“I think people are starting to see how much of an impact we’re having. I mean, we helped flip the Senate,” Nurse told CBC Sports, referring to the work players did to encourage Georgians to vote specifically for Democrat Raphael Warnock and against Republican Sen. Kelly Loeffler, the former owner of the WNBA’s Atlanta Dream who spoke out against Black Lives Matter.

The all-women’s broadcast could have the same kind of impact, said Nurse, the 25-year-old from Hamilton, Ont., who played last season with the WNBA’s New York Liberty and was recently traded to the Phoenix Mercury. 

“You have women who are doing an incredible job across different industries and different nations … coming together to show you guys what we’ve been working on, even though it hasn’t been in the spotlight.”


Nurse will work as a colour analyst for the game alongside play-by-play woman Meghan McPeak, who works for CBC Sports as well as in the booth for the WNBA’s Washington Mystics and the G League’s Capital City Go-Go.

TSN’s Kayla Grey will handle sideline reporting duties, while SportsCentre host Kate Beirness and Amy Audibert, an analyst for Raptors 905, will pair up for the in-game studio show.

Paving the way for the next generation

Nurse, who works as a TSN analyst during the WNBA off-season, says the broadcast will provide a template for young women to see what’s possible in a male-dominated sport and media industry.

“It wasn’t until I went to the U.S. one day and saw Maya Moore on television, I thought, ‘Oh, this would be cool to play in like a national championship and to play at UConn and whatnot,’ ” Nurse said.


CBC Sports contributor Meghan McPeak will call the March 24 Raptors game against the Nuggets as part of TSN’s all-female broadcast crew. (Nathan Denette/The Canadian Press)

McPeak echoed the sentiment on a recent episode of North Courts, a CBC Sports basketball show.

“That’s something that I never had growing up, so the fact that myself, Kia and Kayla can give that to little Black girls that look like us, that’s a fantastic feeling,” McPeak said. 

“Representation matters and little girls will be able to see us doing what we do and might think that they can do it as well.”

As the women call an important Raptors game against a top-tier team the night before the NBA trade deadline, Nurse hopes it will provide a platform to continue speaking out in the name of change — just as she did last summer in the WNBA.

WATCH | McPeak discusses historic broadcast on North Courts:

It’s March Madness time and with a record number of Canadians in the NCAA tournament, we’re dedicating this episode to the stars from north of the border, including Jevohn catching up with Gonzaga’s own sixth man of the year Andrew Nembhard. 17:17

“We took to the court with the understanding that no matter what anybody was going to say about us … some people were going to like what we had to say and some people weren’t,” Nurse said. 

“There’s a fine line between right and wrong. And we knew what was right.”

Providing inspiration for NCAA women’s athletes

To that end, after players in the NCAA women’s March Madness tournament used social media to expose how inferior their weight room setup was compared to the men’s teams, NCAA staff revamped the underwhelming setup with more equipment and machines.  

Other inequities, such as unequal COVID-19 testing and a lack of camera exposure, are also coming to light.

Nurse was part of four Final Four teams with the University of Connecticut, and though she says she never had an issue with weight rooms specifically, she also never played the tournament in a bubble due to a pandemic. She said she’s confused as to how the unequal set-up happened, but isn’t surprised that it did.

“I’m proud of the young women who are at the tournament who took to social media and stood up for themselves,” Nurse said.

“I’m proud of everybody who rallied around them and continued to make it loud enough that the NCAA listened right away. But it shouldn’t have happened in the first place.”



Nurse hopes that she and her fellow WNBA players can provide inspiration for today’s college athletes to continue to find their own voices and feel empowered to speak out. 

“Because they know that if they’re working toward a league like the WNBA, then they’ll still have a voice when they get there.”

Meanwhile, Canadian national team head coach Lisa Thomaidis said television broadcasts like Wednesday’s should become the norm.

“All these steps along the way, they’re massive, right? They shouldn’t be, but they are,” she said. “The fact that we’re going to have an all-female broadcast crew just speaks to how far we’ve come.” 

A role model on the court and in the booth

Nurse spoke to CBC Sports as part of her partnership with Tangerine Bank, which committed $ 15,000 to support Kia Nurse Elite, her Nike-backed youth basketball program.

When she was younger, Nurse said scheduling didn’t allow her to play provincially, nationally, and with the Amateur Athletic Union (AAU) in the U.S. all at once. With the Elite program, she’s aiming to change that while providing young Canadian female basketball players the support they need.

“Every door that I’ve had opened up in my life has been a direct result of being able to play basketball and to playing at a high level in the community that I was a part of growing up.” 

As a player, Nurse is used to serving as a role model for young Canadian basketball fans. On Wednesday, she’ll continue to influence the next generation — but this time, from the booth.

“Hopefully, if there are young women who are watching the game with their families, which I’m sure they are, seeing more people that look like them, maybe one of us resonates with them. And that’s all that matters in this case.”

Let’s block ads! (Why?)

CBC | Sports News

‘Build it and they will come:’ How a Black Canadian coach inspired a generation of hockey players

Cyril Bollers’ ultimate goal in coaching is to reach the NHL. But for now, he’s happy leading Team Jamaica. 

“I think there’s been a lot of frustration in the past with me that I have all the certifications … I just don’t know why I haven’t been given that opportunity,” Bollers said. “But there’s other coaches that are in the same boat of colour that haven’t been given that opportunity either. So I’m not going to say it’s just me, but for me, my goal is to one day coach for Team Canada.”

The 52-year-old doesn’t expect to make a jump straight to the NHL or Olympics, and speculates that the reason he hasn’t advanced much, despite recommendations from the likes of Hockey Hall of Famer Paul Coffey, is a lack of connections at the next level.

“I don’t want to say it’s colour, especially with hockey being for everyone. Other people may — I don’t. I just want to say that the opportunity hasn’t arisen yet and I’m hoping it does. So based on that, I’m continuing to network,” Bollers said.

Bollers is the president of Skillz Black Aces, a Toronto-based program that helps bring hockey to underprivileged and BIPOC youth. It has produced NHLers such as Anson Carter, Wayne Simmonds and brothers Anthony and Chris Stewart.

Born in Guyana, Bollers now lives in Scarborough, Ont., after moving to Canada when he was four. He was inspired to become a coach when his son was six and playing house-league hockey for a coach who heavily favoured his own son.

Soon, coaching became a passion. He’s since worked with the Toronto Red Wings and Marlboros of the GTHL and the Pickering Panthers of the OJHL.

“I was told that I couldn’t because of the colour of my skin, which fuelled the fire, which promoted the education in regards to quality certificates, which gave me the opportunity to prove others wrong,” Bollers said.

In addition to his work with the Black Aces, Bollers has also served for the past four years as head coach and general manager of Team Jamaica — a country that doesn’t contain so much as a single ice rink.

Bollers also works with the Black Canadian Coaches Association in hopes of reaching a broader base of BIPOC coaches throughout the country to serve as a mentor and to help create a network between coaches and sports organizations.

Legacy with Black Aces

But it was with the Black Aces where Bollers helped inspire a generation of BIPOC players, many of whom followed him to Team Jamaica.

“I guess when they say build it and they will come, that’s what it was. Everybody wanted to become a Skillz Black Ace,” Bollers said.

The program began around 20 years ago, partially the brainchild of former NHL goalie Kevin Weekes, as a camp that would run a few times per year. Bollers helped build it into more of a team that would enter — and quickly dominate — tournaments against top competition.


A Skillz Black Aces team is seen above at a tournament. (Courtesy Cyril Bollers)

In addition to a heavy majority of BIPOC players, Bollers led a group of five Black coaches on the bench. The team consistently stunned its opponents with blazing speed and won way more often than it lost.

For parents of colour, the Black Aces was an opportunity to show their children there are other hockey players who look like them.

“And that was the main thing was he was not an outsider or ‘that one kid’ with this group,” said Mark Francis, whose son Peyton played for the Aces and now plays centre for the University of Alabama-Huntsville Chargers.

Loren Francis heard racist comments from the stands when she watched her son play on predominantly white teams. Since Loren is white, other parents did not realize she was Peyton’s mom. When the Black Aces opportunity arose, Mark and Loren were intrigued.

“I thought this was going to be more like a how-to-play hockey type of thing,” Mark said. “And then we went out and I was shocked because not only were the kids very highly skilled, but [Bollers’] coaching methods, I would say, were top notch.”


Bollers leads a player through a drill. (Courtesy Cyril Bollers)

Vancouver Canucks forward Justin Bailey is another Black Aces alumnus. Born in Buffalo, N.Y., a 12-year-old Bailey was hesitant about joining a team across the border where he didn’t know anyone.

It took some convincing from his mother, Karen Buscaglia, and the decision was an instant success.

“People embraced their differences. And they had fun music playing in the locker room. And it was the first time that I could look at him and I could see he just had a blast. And obviously hockey was predominantly white, so he had never been exposed to anything like that,” Buscaglia said.

While a fun atmosphere certainly existed around the Black Aces, both Francis and Buscaglia say Bollers ran a tight ship where discipline among players — things like walking in an orderly fashion and politeness — impacted players’ ice-time.

The Black Aces, counting one edition of the team featuring one of Bollers’ three sons, often faced racism from other teams, including hearing the N-word uttered against them on the ice.

“We used to laugh at it because we were so good we would beat people. And for me, I would just tell the guys, ‘They can’t beat you on the ice. They’re going to try to beat you with their words. But words are just words,'” Bollers said.

Equal success with Jamaica

As a white player born in the Caribbean, Ethan Finlason had a slightly different experience when he joined Bollers’ Team Jamaica. Finlason played inline hockey in his home country of the Cayman Islands before eventually moving to Canada to pursue ice hockey.

He was met with hostility from other kids who said he should quit because he was Caribbean. Then a goalie from his academy team stayed behind to watch one of the team’s games.

“The Canadian goalie was shocked that Jamaicans could skate,” said Ethan’s father Andrew. “And I don’t know where this bias comes from. I mean, most of these kids grew up in Canada. But they’re tremendous athletes. They have a tremendous coach. But there’s this stigma that they shouldn’t be able to play.”


Team Jamaica players are seen above in 2014. (Courtesy Cyril Bollers)

In 2019, Jamaica went 5-0 en route to winning the championship at the LATAM Cup, an international tournament pitting top Latin American and Caribbean teams.

But Jamaica can’t be fully sanctioned by the IIHF until it builds a rink. When that happens, more resources could be poured into the program and the pitch to NHL players of Jamaican descent, like the Subbans, can begin.

“I’m sure that once that’s happened, we can just place a call to Karl [Subban] and then Karl will round up the boys and then we’ll take it from there. But I think until it’s fully sanctioned, we don’t want to put the cart before the horse,” Bollers said.

When that finally happens, Bollers said his admittedly lofty goal is to qualify for the Olympics.

Between the Black Aces and Team Jamaica, Bollers’ hands are plenty full in the world of hockey, even as he continues to eye a pro position. He can take solace in the fact that if nothing else, his teams simply win.

“They used to come and watch us play because we were fast, we were strong, it was entertaining hockey. But more importantly we could coach, and I think what people forget is I’m a hockey coach by choice, Black by nature.”

Let’s block ads! (Why?)

CBC | Sports News

Canada’s Sébastien Toutant inspires next generation following 2018 Olympic gold

EDITOR’S NOTE: Relive the action from the men’s big air snowboarding final at Pyeongchang 2018 on Wednesday beginning at 9 a.m. ET in our latest instalment of Rewind Wednesday..

Two years ago, as Canadian snowboarder Sébastien Toutant was throwing down his biggest tricks to capture gold in the inaugural Olympic big air competition in Pyeongchang, South Korea — halfway around the world — two teenage snowboarders pulled into a Tim Hortons near the Quebec-Vermont border to watch the spectacle on their cell phones.

“We needed the WiFi,” laughs Nic Laframboise, now 20 and a member of Canada Snowboard’s national team.

Watching his idol “Seb Toots” that day, alongside friend and fellow Canadian rider Frank Jobin, made him realize his own Olympic dream was in sight.

“All the tricks they were doing, I wasn’t close at that time, but I knew I could do those tricks on a trampoline and I could do the easier version of what they were doing,” he said. “I knew it was possible in the future.”

Perhaps as Canadian as a road-trip stop at Tim Hortons is the tradition of dominance in slopestyle and big air snowboarding in this country.

It could be Mark McMorris, the most decorated snowboarder in the history of the Winter X Games (along with two Olympic medals amid a litany of career-threatening injuries), or Max Parrot, the 2018 Olympic slopestyle silver medallist who beat cancer and came back to win big air at X Games this past February, or the stylish Toutant, another past X Games champion, who made history as the inaugural Olympic big air winner.

WATCH | Seb Toutant claims inaugural Olympic big air gold:

The Canadian soared to new heights and the win puts his name in the record books as Canada’s 500th Olympic medal. 0:43

On the women’s side, Laurie Blouin is a world champion, to go along with silver at the Olympics and X Games.

They’ve got the goods — the tricks, the flair, the cred — and they’re all still near the top of their game. So how does the next generation of riders crack the Canadian team?

Canada Snowboard’s slopestyle head coach Elliot Catton says given how rich in talent the program is, it’s tough.

“There’s no easy way to put it,” he said this week from his home in Squamish, B.C. “We have some of the best in the world, just the standards of making the team are higher than in other disciplines or another sport.

“But it just comes down to the skills they can acquire, the tricks they can do and then being able to do those when it counts — performance on-demand in competitions. That’s what it takes in any sport to be up at the top, but in particular to this team.”

The next generation

Jasmine Baird wants to be on that Canadian team for Beijing 2022, but she has some work to do.

The 21-year-old national team member, originally from Mississauga, Ont., has been sidelined since last August after she tore the ACL in her left knee in Cardrona, New Zealand.

“For about the last 10 years, my goal has been to get to the 2022 Olympics,” she said. “Being put on the sideline has definitely been a struggle mentally for me because everyone else I’ll be competing against has been progressing these past 12 months, but I haven’t.”

Since reconstructive surgery, she’s been rehabbing religiously, working on her strength, mobility and flexibility until she can get back on snow.

Before her injury, she was making noise on the international stage with her first World Cup podium in January 2019 (third in slopestyle at Seiser Alm, Italy) and as the top Canadian in her world championship debut in Park City, Utah (sixth in slopestyle, ahead of Olympians Brooke Voigt and Blouin).


Jasmine Baird, 21, of Mississauga, Ont., has been battling back from a 2019 torn ACL injury with the goal of claiming a spot in Beijing 2022. (Canada Snowboard)

She remembers watching the Pyeongchang big air finals and comparing her skills to what was being put down at the Olympics.

“Some of them I’d be like ‘oh, I can do that trick’ so it was a good feeling and confirmation that I was on the right path,” she said. “It definitely made me feel confident about where I was at with my progression and my skills looking at 2022.”

She says her biggest trick is a cab double 900, which is a switch double underflip. She learned it last season before getting injured and kindly walked this author through the move.

“You go in switch, so your unnatural way of riding, and you take off 90 degrees off the takeoff, then double backflip and continue with another 90 degree rotation to land regular, on your natural stance of riding. So it’s a 900 degree trick by the end.”

Just to put in context: the biggest trick in Pyeongchang was a 1080, done by gold medallist Anna Gasser of Austria. So to be in the mix with the top women in the world in big air, Baird hopes to add a 10 to her repertoire this season.

Laframboise, or “The Flying Raspberry” as he’s called, put his name in the conversation for Beijing 2022 after finishing second overall in the World Cup big air points standings, which included his first victory in Modena, Italy.

“I did not expect it,” he said from his home in Saint-Jean-sur-Richelieu, Que. “I knew I was capable of it, but being capable and actually doing it is two different things.”

He said the biggest lessons he learned from his first full season on the senior team are performing under pressure and working on his slopestyle game.


Nic Laframboise’s impressive run of performances earned him second overall in the World Cup big air points standings. A great achievement for the young rider in his first full season with the senior Canadian snowboarding team. (Canada Snowboard)

Coach Catton says that’s something the younger riders learn as they grow in their careers.

“Big air is focusing on one jump at a time versus linking rail tricks and your jump tricks together,” Catton said. “Nic showed in big air that he has the ability, he just needs to work on developing a little bit of a different skill-set in slopestyle. He’s got it all there. It’s just something that comes with time and experience. He’s shown a lot of promise and we’re really stoked for him on his results and his riding in the big air part of the season.”

Another bonus for Laframboise this season was getting to pick the brains of champions like Toutant and Parrot, who he got to room with in Laax, Switzerland, and at the US Open in Vail, Colo., respectively.

“Every time I snowboard with Seb it’s super fun. I think he has a great way of seeing snowboarding and making it fun. I like that attitude. And with Max, he’s really focused and in the zone. I got to ask him about how he chills out before contests.”

Every little edge counts on the road to Beijing, especially with the crop of veteran names and those coming up alongside Laframboise, like three-time Youth Olympic Games medallist Liam Brearly and double world junior medallist Will Buffey, among others.

Progression between Olympics

Beijing 2022 is less than two years away and this should be the year when riders start testing out new tricks and pushing the standards from snowboarding’s past.

“I think the sport is constantly evolving and progressing. That’s kind of the nature of it,” Catton said. “Big tricks are great, but within snowboarding there’s different grabs and spinning on a different axis that can help make things unique. So it’s not just another 180 or another flip that’s going to be the next big thing. There’s a creative aspect.”

Toutant won gold with a backside 1620 (turning backward on takeoff, four and a half rotations and flipping off-axis three times) and frontside triple-cork 1440 (going forward off the jump, four full rotations and three off-axis spins), both still huge tricks to this day.

Many of the top men on tour do 1620s and 1800s, while the top women are doing 1080s and some are doing 1260s.

One other thing the riders, coaches and fans are looking forward to in Beijing is the venue for big air. The brand-new Shougang Park permanent big air ramp, located on the western outskirts of China’s capital city, was on display at the Air & Style event in early Dec. 2019.


The big air ramp at Shougang Park, seen above during the Air + Style competition in 2019, is a new venue which will feature in the Beijing 2022 Winter Games. (Lintao Zhang/Getty Images)

As for which countries will dominate come Beijing 2022, in terms of big air and slopestyle, especially on the men’s side, it’s Canada, says Craig McMorris. The professional snowboarder has watched the sport over the past two Olympic Winter Games in his role as CBC’s colour commentator.

“Just because of the depth. At every level, Canada has three or four guys who could be the No. 1 guy. The 16- and 17-year-olds, the 21- and 22-year-olds and the 25 and ups. Every country has some really good riders, but Canada has three or four great riders in every single age category.

“I feel like qualifying for the Olympics on Team Canada is harder almost than going to the Olympics itself. At the Olympics, you get one day. If you’re on, you’re on. But the qualification to even get on to Team Canada is all season, every season. I think that’s one of the hardest things to do in sports, to be honest.”

For the women in big air, McMorris says it’s early, but watch out for the Japanese.

“They have three or four riders right now doing insane tricks and when they get a little more consistent, it’s not outlandish to say they’d sweep the podium.”

Relive the action from the dramatic final beginning on Wednesday at 9 a.m. ET on CBC and CBCSports.ca.

Let’s block ads! (Why?)

CBC | Sports News

Hip and sober: the mocktail generation

Toronto bartender Joel Savoie sets his ingredients on the bar and mixes the first of many drinks he’ll prepare for his patrons on a muggy summer evening.

The drink, a golden concoction called spiced tepache, is mostly fermented pineapple husk. The spicy part comes from a dried-up worm, the kind you find at the bottom of a bottle of mescal. It’s used to salt the rim. Missing from the ingredients is alcohol. There’s not a drop of it in the glass.

“We will generally sell more alcoholic drinks than non-alcoholic drinks,” says Lavoie. “But people who come here for placebos are excited.”

The bar, called PrettyUgly, is one of many drinking establishments in Toronto offering a selection of mocktails, or placebo cocktails, as they’re called here.

“The whole idea about the placebo program is to integrate those drinks within the menu so people won’t be pointed out as ‘OMG, you’re not drinking’. Because all of a sudden that segregates that particular clientele,” says Evelyn Chick, the bar’s general manager. So far the bar is offering six non-alcoholic drinks. 


Bars are increasingly catering to patrons who are abstaining from alcohol, says PrettyUgly’s general manager Evelyn Chick. The Toronto bar serves placebo cocktails as an alternative to the real deal. (Craig Chivers/CBC)

Sobriety movement is growing

The World Health Organization says 2.3 billion people are current drinkers and global consumption is expected to increase in the next 10 years.

And a recent study in the Canadian Medical Journal offered some sobering statistics on emergency department visits in Ontario hospitals. It showed a dramatic rise in visits because of alcohol, especially by young women.


As imbibing rates increase across the world, some bars in Canada have created non-alcoholic drinks for abstaining patrons. Campaigns like Dry January and Sober Curious encourage people to take a break from alcohol. (Craig Chivers/CBC)

In the face of a growing imbibing culture, some young adults are going sober, or at least cutting back on the booze. On social media, established campaigns like Dry July and Sober October extol the virtues of abstinence. Campaigns like Mindful Drinking and Sober Curious have also created popular catchphrases. 

“People are always trying to find ways to resist alcohol and keep it out of their lives or minimize harms,” says Prof. Tim Stockwell. He’s the director of the Canadian Institute for Substance Use Research at the University of Victoria. 

Can’t afford to get drunk

“There’s a trend around the world of younger people drinking less and more people abstaining,” he says. Cultural trends like Sober Curious and Mindful Drinking are a reflection of that. Stockwell doesn’t believe the campaigns are a fleeting fad.

“They really have influenced the culture and created more opportunities and made it more possible for people to choose to be abstinent.”

Stockwell believes the driving force behind this is economics. Young adults have less disposable income and are “up to their ears in debt. They can’t afford to to be out of their brains anymore,” he says.


Ryla Parker and Graham Gibb are social drinkers but are trying out “placebo” cocktails at a Toronto bar. Teetotaller nights and non-alcoholic drinks are gaining popularity as more young adults embrace sobriety. (Craig Chivers/CBC )

The internet is changing everyone’s behaviour, he adds, creating an environment where people are more mindful and careful in what they say and do and try to be more in control of themselves.

Even abstaining from alcohol for a short period has health benefits, he says. “There’s evidence people sleep better. Their mental well-being improves. They take more exercise. They look after themselves better, and they tend to be happier.”

At PrettyUgly, Ryla Parker and her friend Graham Gibb have popped in after work. The 20-somethings are self-proclaimed social drinkers but are giving placebo drinks a shot.

“You don’t always want to get drunk and have a hangover the next day,” says Parker. Gibb agrees. “It’s nice to get out and drink after work without feeling the effects the next day.”

Let’s block ads! (Why?)

CBC | Health News

How to Create Your Own State-of-the-Art Text Generation System

Hardly a day goes by when there isn’t a story about fake news. It reminds me of a quote from the favorite radio newsman from my youth, “If you don’t like the news, go out and make some of your own.” OpenAI’s breakthrough language model, the 1.5 billion parameter version of GPT-2, got close enough that the group decided it was too dangerous to release publicly, at least for now. However, OpenAI has now released two smaller versions of the model, along with tools for fine-tuning them on your own text. So, without too much effort, and using dramatically less GPU time than it would take to train from scratch, you can create a tuned version of GPT-2 that will be able to generate text in the style you give it, or even start to answer questions similar to ones you train it with.

What Makes GPT-2 Special

GPT-2 (Generative Pre-Trained Transformer version 2) is based on a version of the very powerful Transformer Attention-based Neural Network. What got the researchers at OpenAI so excited about it was finding that it could address a number of language tasks without being directly trained on them. Once pre-trained with its massive corpus of Reddit data and given the proper prompts, it did a passable job of answering questions and translating languages. It certainly isn’t anything like Watson as far as semantic knowledge, but this type of unsupervised learning is particularly exciting because it removes much of the time and expense needed to label data for supervised learning.

Overview of Working With GPT-2

For such a powerful tool, the process of working with GPT-2 is thankfully fairly simple, as long as you are at least a little familiar with Tensorflow. Most of the tutorials I’ve found also rely on Python, so having at least a basic knowledge of programming in Python or a similar language is very helpful. Currently, OpenAI has released two pre-trained versions of GPT-2. One (117M) has 117 million parameters, while the other (345M) has 345 million. As you might expect the larger version requires more GPU memory and takes longer to train. You can train either on your CPU, but it is going to be really slow.

The first step is downloading one or both of the models. Fortunately, most of the tutorials, including the ones we’ll walk you through below, have Python code to do that for you. Once downloaded, you can run the pre-trained model either to generate text automatically or in response to a prompt you provide. But there is also code that lets you build on the pre-trained model by fine-tuning it on a data source of your choice. Once you’ve tuned your model to your satisfaction, then it’s simply a matter of running it and providing suitable prompts.

Working with GPT-2 On Your Local Machine

There are a number of tutorials on this, but my favorite is by Max Woolf. In fact, until the OpenAI release, I was working with his text-generating RNN, which he borrowed from for his GPT-2 work. He’s provided a full package on GitHub for downloading, tuning, and running a GPT-2 based model. You can even snag it directly as a package from PyPl. The readme walks you through the entire process, with some suggestions on how to tweak various parameters. If you happen to have a massive GPU handy, this is a great approach, but since the 345M model needs most of a 16GB GPU for training or tuning, you may need to turn to a cloud GPU.

Working with GPT-2 for Free Using Google’s Colab

I kept checkpoints of my model every 15,000 steps for comparison and in case the model eventually overfit and I needed to go back to an earlier version.Fortunately, there is a way to use a powerful GPU in the cloud for free — Google’s Colab. It isn’t as flexible as an actual Google Compute Engine account, and you have to reload everything each session, but did I mention it’s free? In my testing, I got either a Tesla T4 or a K80 GPU when I initialized a notebook, either one of which is fast enough to train these models at a reasonable clip. The best part is that Woolf has already authored a Colab notebook that echoes the local Python code version of gpt2-simple. Much like the desktop version, you can simply follow along, or tweak parameters to experiment. There is some added complexity in getting the data in and out of Colab, but the notebook will walk you through that as well.

Getting Data for Your Project

Now that powerful language models have been released onto the web, and tutorials abound on how to use them, the hardest part of your project might be creating the dataset you want to use for tuning. If you want to replicate the experiments of others by having it generate Shakespeare or write Star Trek dialog, you can simply snag one that is online. In my case, I wanted to see how the models would do when asked to generate articles like those found on ExtremeTech. I had access to a back catalog of over 12,000 articles from the last 10 years. So I was able to put them together into a text file, and use it as the basis for fine-tuning.

If you have other ambitions that include mimicking a website, scraping is certainly an alternative. There are some sophisticated services like ParseHub, but they are limited unless you pay for a commercial plan. I have found the Chrome Extension Webscraper.io to be flexible enough for many applications, and it’s fast and free. One big cautionary note is to pay attention to Terms of Service for whatever website you’re thinking of, as well as any copyright issues. From looking at the output of various language models, they certainly aren’t taught to not plagiarize.

So, Can It Do Tech Journalism?

Once I had my corpus of 12,000 ExtremeTech articles, I started by trying to train the simplified GPT-2 on my desktop’s Nvidia 1080 GPU. Unfortunately, the GPU’s 8GB of RAM wasn’t enough. So I switched to training the 117M model on my 4-core i7. It wasn’t insanely terrible, but it would have taken over a week to make a real dent even with the smaller of the two models. So I switched to Colab and the 345M model. The training was much, much, faster, but needing to deal with session resets and the unpredictability of which GPU I’d get for each session was annoying.

Upgrading to Google’s Compute Engine

After that, I bit the bullet, signed up for a Google Compute Engine account, and decided to take advantage of the $ 300 credit Google gives new customers. If you’re not familiar with setting up a VM in the cloud it can be a bit daunting, but there are lots of online guides. It’s simplest if you start with one of the pre-configured VMs that already has Tensorflow installed. I picked a Linux version with 4 vCPUs. Even though my desktop system is Windows, the same Python code ran perfectly on both. You then need to add a GPU, which in my case took a request to Google support for permission. I assume that is because GPU-equipped machines are more expensive and less flexible than CPU-only machines, so they have some type of vetting process. It only took a couple of hours, and I was able to launch a VM with a Tesla T4. When I first logged in (using the built-in SSH) it reminded me that I needed to install Nvidia drivers for the T4, and gave me the command I needed.

Next, you need is to set up a file transfer client like WinSCP, and get started working with your model. Once you upload your code and data, create a Python virtual environment (optional), and load up the needed packages, you can proceed the same way you did on your desktop. I trained my model in increments of 15,000 steps and downloaded the model checkpoints each time, so I’d have them for reference. That can be particularly important if you have a small training dataset, as too much training can cause your model to over-fit and actually get worse. So having checkpoints you can return to is valuable.

Speaking of checkpoints, like the models, they’re large. So you’ll probably want to add a disk to your VM. By having the disk separate, you can always use it for other projects. The process for automatically mounting it is a bit annoying (it seems like it could be a checkbox, but it’s not). Fortunately, you only have to do it once. After I had my VM up and running with the needed code, model, and training data, I let it loose. The T4 was able to run about one step every 1.5 seconds. The VM I’d configured cost about $ 25/day (remember that VMs don’t turn themselves off; you need to shut them down if you don’t want to be billed, and persistent disk keeps getting billed even then).

To save some money, I transferred the model checkpoints (as a .zip file) back to my desktop. I could then shut down the VM (saving a buck or two an hour), and interact with the model locally. You get the same output either way because the model and checkpoint are identical. The traditional way to evaluate the success of your training is to hold out a portion of your training data as a validation set. If the loss continues to decrease but accuracy (which you get by computing the loss when you run your model on the data you’ve held out for validation) decreases, it is likely you’ve started to over-fit your data and your model is simply “memorizing” your input and feeding it back to you. That reduces its ability to deal with new information.

Here’s the Beef: Some Sample Outputs After Days of Training

After experimenting on various types of prompts, I settled on feeding the model (which I’ve nicknamed The Oracle) the first sentences of actual ExtremeTech articles and seeing what it came up with. After 48 hours (106,000 steps in this case) of training on a T4, here is an example:

Output of our model after two days of training on a T4 when fed the first sentence of Ryan Whitwam's Titan article.

The output of our model after two days of training on a T4 when fed the first sentence of Ryan Whitwam’s Titan article. Obviously, it’s not going to fool anyone, but the model is starting to do a decent job of linking similar concepts together at this point.

The more information the model has about a topic, the more it starts to generate plausible text. We write about Windows Update a lot, so I figured I’d let the model give it a try:

The model's response to a prompt about Windows Update after a couple days of training.

The model’s response to a prompt about Windows Update after a couple of days of training.

With something as subjective as text generation, it is hard to know how far to go with training a model. That’s particularly true because each time a prompt is submitted, you’ll get a different response. If you want to get some plausible or amusing answers, your best bet is to generate several samples for each prompt and look through them yourself. In the case of the Windows Update prompt, we fed the model the same prompt after another few hours of training, and it looked like the extra work might have been helpful:

After another few hours of training here is the best of the samples when given the same prompt about Microsoft Windows.

After another few hours of training, here is the best of the samples when given the same prompt about Microsoft Windows.

Here’s Why Unsupervised Models are So Cool

I was impressed, but not blown away, by the raw predictive performance of GPT-2 (at least the public version) compared with simpler solutions like textgenrnn. What I didn’t catch on to until later was the versatility. GPT-2 is general purpose enough that it can address a wide variety of use cases. For example, if you give it pairs of French and English sentences as a prompt, followed by only a French sentence, it does a plausible job of generating translations. Or if you give it question-and-answer pairs, followed by a question, it does a decent job of coming up with a plausible answer. If you generate some interesting text or articles, please consider sharing, as this is definitely a learning experience for all of us.

Now Read:

Let’s block ads! (Why?)

ExtremeTechExtreme – ExtremeTech

Intel Unleashes 9th Generation 8-Core Mobile CPUs

This site may earn affiliate commissions from the links on this page. Terms of use.

Intel is announcing its new, 9th generation laptop processors today, in a surprisingly competitive strike for a company not generally known for dramatic product shuffles.

From 2011 to early 2017, Intel stuck to essentially the same price bands and SKU structure for its desktop and laptop processors. Standard desktop chips ranged from dual-cores without Hyper-Threading at the bottom of the stack to quad-core chips with HT at the top. Laptops were almost entirely dual-core, with a handful of quad-core parts. This only began to change in early 2017, as a response to AMD’s incipient Ryzen CPU. But while Intel has expanded both desktop and laptop CPU core counts, it’s been taking these moves in reaction to what AMD was doing — not trying to get out ahead.

Intel’s 9th Generation mobile CPU parts are an obvious attempt to put the company back on a more aggressive footing. Intel’s Core i9-9980HK has a single-core turbo frequency of up to 5GHz and a whopping eight-core/16-thread configuration. Intel claims it can deliver the chip in the same 45W TDP envelope as the Core i9-8950HK. For comparison, that CPU had a base clock of 2.9GHz and a boost frequency of 4.8GHz. The Core i9-9980HK has a base frequency of 2.4GHz and the aforementioned 5GHz boost. The chip will also feature Intel’s Thermal Velocity Boost, which allows a CPU to run as much as 200MHz faster in short bursts if the Tcase temperature is below 50C.

The slideshow below summarizes Intel’s new CPUs, increased performance expectations, and other features of the platform. As expected, Intel is positioning the H10 Memory solution that combines QLC NAND with an Optane cache drive.

The full suite of 9th Generation CPUs launching today is shown below. All of them share a 45W TDP. This likely translates to higher sustained frequencies for the lower core count quad-cores.

9th-Gen-Clock-Rates

Click to enlarge.

When Intel revamped the 8th generation family and introduced quad-core mobile chips, it dramatically cut the base clock on these CPUs.SEEAMAZON_ET_135 See Amazon ET commerce At the time, we warned that this could have performance impacts if OEMs didn’t cool systems appropriately — a fast dual-core can outperform a quad-core if the latter is severely throttled. This problem, however, never materialized. 8th gen quad-core mobile chips were significantly faster than the dual cores they replaced.

The same should be true here as well, though obviously not to quite the same degree. Moving from two cores to four doubled the available core count, while the four-to-six shift is a 1.5x improvement and the six-to-eight step is just 1.33x. Mobile owners should be aware that with higher core counts comes more heat at the same clock speed, and that these chips probably won’t run as fast as lower core counts — but with 8th Gen, Intel demonstrated it could calibrate its chips to deliver impressive gains. Assuming OEMs do their job with proper cooling, these CPUs should deliver gains over their predecessors, even in the limited confines of laptop cooling.

We don’t necessarily expect a near-term response from AMD. Third-generation Ryzen will launch in the coming months, with Navi and undoubtedly a Ryzen Mobile refresh at some point after that. The shift to 7nm will undoubtedly give AMD new options for its APUs. The only thing the company has said to date, however, is that we should not expect a Matisse-style APU (this would be an APU with both a CPU and a GPU chiplet).

We also don’t know where AMD will focus its efforts. The company could challenge Intel across the entire mobile market, or it might choose to focus on low power bands, emphasizing improving its competitive standing in the areas where 7nm has the most to offer. Either way, Intel’s decision to launch an eight-core mobile chip means it’s serious about driving the performance conversation.

Now Read: 

Let’s block ads! (Why?)

ExtremeTechExtreme – ExtremeTech

Intel’s New Core i5-8250U is a Huge Upgrade Over Older 7th Generation CPUs

This site may earn affiliate commissions from the links on this page. Terms of use.

When Intel launched its new 8th generation quad-core processors in August, it wasn’t clear what kind of an upgrade we were actually getting. While putting a quad-core CPU in a modern laptop at 15W was a major achievement, there was always a question of how fast it would run or what kind of performance improvement users could expect.

The answer, according to Tech Report? Lots. Lots and lots. In fact, the Core i5-8250 delivers an average performance uplift in CPU tasks of 59 percent. That’s an enormous improvement in the 15W form factor, and Intel has pulled it off with apparently little impact on battery life.

Tech Report put a Core i5-8250U laptop up against the older dual-core Core i5-7200U and a 35W i7-7700HQ. While the 45W quad-core CPU is obviously going to be faster in most respects, the Core i5-8250 is far and away the better CPU over the Core i5-7200U.

cinent

Obviously benchmark results are going to vary from manufacturer to manufacturer. Intel now gives OEMs more freedom than it used to when it comes to picking the skin temperatures or other features that they want to target, which means performance can vary from laptop to laptop. But the overall gains here are surprisingly good — honestly better than I would’ve expected.

Of course, this is a move from Intel to preempt Raven Ridge and Ryzen from blowing holes in their mobile product lines the way Ryzen and Threadripper punched through Kaby Lake and the X-Series family earlier this year. Still, consumers who pick up 8th generation quad-core mobile chips can look forward to a vastly better experience than seventh generation CPUs offered, without a huge leap in price.

If you’re a gamer you’ll want to check the game benchmarks as well. The long and short of it is that the 8th generation UHD 620 graphics are moderately faster than the old HD 620, but not overwhelmingly so. Gamers will still want to find a dedicated GPU alternative.

Hopefully the advent of quad-core laptops, even in the 15W space, will speed the adoption of multi-core CPUs and encourage game developers to spend more time optimizing APIs like DirectX 12. The desktop/laptop split is something like 30/70 these days, meaning dual-core configurations have to factor into developer targets. With more CPU cores to play with, it makes more sense to optimize for those scenarios, and we might finally start to see some gains from DX12 as a result.

Let’s block ads! (Why?)

ExtremeTechExtreme – ExtremeTech

Stepping back but not down: How the Queen is gradually shifting duties to the next generation

For a Queen deeply devoted to duty who served in the Second World War, laying a wreath of remembrance has always been a significant and solemn occasion.

But this year is different.

Instead of Elizabeth carefully stepping up the steps of the cenotaph in central London on Sunday during the annual remembrance ceremony, her eldest son and heir, Prince Charles, is to take on her wreath-laying responsibility. The 91-year-old monarch will watch from a nearby balcony.

That simple shift represents one of the most obvious and high-profile signals yet of the gradual transition unfolding in the upper echelons of the House of Windsor, now headed by the longest-reigning British monarch.

(FILE) BRITAIN ROYALTY REMEMBRANCE SUNDAY

Elizabeth lays a wreath during the Remembrance Sunday ceremony at the cenotaph in London on Nov. 13, 2016. (Will Oliver/EPA-EFE)

“While Remembrance Sunday may be the most dramatic evidence of the shifts in the Royal Family, it’s very much on a continuum of small and steady changes, which is really what the Queen’s reign has been about,” says author Sally Bedell Smith, who has written biographies of Elizabeth and Charles.

Gradually over the past few years, some duties carried out by the Queen have been taken on by the younger generations of the Royal Family, including overseas travel and some investitures.

Earlier this year, her husband, Prince Philip, the 96-year-old Duke of Edinburgh, stepped back from official duties. But he sometimes appears in public, and he’s to be on the balcony where other family members traditionally watch the ceremony on Sunday.

Just like Philip

“He had already decided not to lay a wreath at the cenotaph, and her decision was in line with his,” says Smith.

“The bottom line is that the presence of the Queen and Prince Philip as observers rather than participants on Remembrance Sunday will underline the transition that has been occurring in the Royal Family since before the Queen’s 90th birthday.”

Some who watch the Royal Family closely see practicality and careful consideration for the Queen’s age as critical factors in the decision to watch rather than lay a wreath.

BRITAIN-POLITICS/

When Queen Elizabeth attending the state opening of Parliament in London on June 21, 2017, the 1.4-kilogram Imperial State Crown was carried in rather than being worn by the Queen. (Stefan Rousseau/Reuters)

“I think the remarkable thing is that she kept going for so long,” says Ingrid Seward, editor in chief of Majesty Magazine and whose book My Husband and I, on the Queen and Prince Philip’s 70-year marriage, was published this month.

Now, there is “relief all round” with the fact Elizabeth and Philip will be watching rather walking up the cenotaph steps for the Remembrance Sunday ceremony, “as they could have fallen or anything like that and then been in trouble,” says Seward.

Other changes have been made as the Queen grows older.

“She now takes shorter routes and climbs fewer steps when she goes on walkabouts,” says Smith.

No heavy robes or crown

She also did not wear ceremonial robes or the elaborate — but weighty — Imperial State Crown to the opening of Parliament earlier this year.

“Since the crown weighs three pounds [1.4 kilograms], and the robes are heavy, that could well be the routine for the Queen in years ahead,” says Smith.

In as much of some of her duties have shifted and accommodations have been made, it seems Elizabeth’s hand is still firmly on the royal tiller.

“She’s still holding investitures, just not as many, she’s still taking part in receptions and hosting receptions, meeting people,” says Penny Junor, whose book on Charles’s second wife, Camilla, the Duchess of Cornwall, was published a few months ago.

The Queen’s activities, Junor suggests, are a tiring prospect for her, shaking hands, keeping a smile on her face and remembering who she’s meeting and why at every reception or garden party.

BRITAIN-ROYALS/

Queen Elizabeth is still holding events and investitures, just not as many as she once did. (Hannah McKay/Reuters)

“I’m not suggesting she’s losing her mind by any stretch of the imagination,” says Junor. “I’m just saying what she does is very tiring and it’s very full on …. in her 90s it’s pretty reasonable for her to be sharing the load.”

That load has spread notably to Charles, with his sister Anne and elder son Prince William also taking on more investitures. Others such as Charles’s brother Edward and his wife, Sophie, the Earl and Countess of Wessex, have also taken on more activities. William gave up his job as an air ambulance pilot earlier this year to focus full-time on royal duties.

As much as the Queen is sharing more of the load, there is no sense that she would consider abdication, as has happened in other royal houses in Europe in recent years.

‘Lifelong commitment’

“The Queen appears to view her position as a lifelong commitment to her people,” says Carolyn Harris, a Toronto-based royal author and historian whose book Raising Royalty: 1,000 Years of Royal Parenting was released earlier this year.

“At the age of 21, she stated that she would devote her whole life whether it was long or short to the service of her people and she received a religious coronation ceremony in 1953 that emphasized this lifelong commitment.”

Plus, abdication casts an unwelcome shadow for Elizabeth.

“For the Queen, abdication is associated with her uncle, Edward the Eighth, and the abdication crisis of 1936, which was very destabilizing for the monarchy because of the Duke and Duchess of Windsor,” says Harris.

CANADA-POLITICS/150

The Queen has given up long-haul overseas travel. Prince Charles and Camilla, Duchess of Cornwall, were in Canada to mark the 150th anniversary since Confederation and visited Parliament Hill in Ottawa for the celebrations on July 1, 2017. (Blair Gable/Reuters)

Still, the passage of time is inevitable, and every reign is succeeded by another.

In considering the next reign, Elizabeth appears to charting this gradual course quite deliberately. And it’s in marked contrast to what happened with other long-serving monarchs such as George III or Victoria.

“In the past, we’ve seen either an abrupt end to being in the public eye in the case of George the Third or a determination not to have the heir to take on too many royal duties in the case of Queen Victoria,” says Harris.

That abrupt end of public visibility for George III — the subject of the film The Madness of King George — came with the declaration of a regency in 1811.

BRITAIN-ROYALS/

Queen Elizabeth was out and about the week before the Remembrance Sunday ceremony, attending the reopening of the Sir Joseph Hotung Gallery of China and South Asia at the British Museum in London on Nov. 8, 2017. (Daniel Leal-Olivas/Reuters)

“All these arrangements are being made with the blessing of the Queen, who is above all pragmatic about the need to hand over duties to her eldest son as she steps back — but emphatically not down,” says Smith.

Junor says she thinks Elizabeth is also very content with the succession.

“I think she’s in a very good place. I don’t think that’s been true over the years because I think she’s probably been anxious about what would happen if she was to die because Charles had been so unpopular and there was the Camilla question and William didn’t look 100 per cent sold on being a royal prince, but all these anxieties have gone.”

Let’s block ads! (Why?)

CBC | World News