Animals That Start with A - Listed With Pictures, Facts
The VERY Bizarre (and Scary) Side Effects Of Ozempic: From Insanely Lucid Dreams About Matthew McConaughey Riding A Llama To Spontaneous Phobias Of Ground Beef, Sagging Skin ...
The VERY bizarre (and scary) side effects of Ozempic: From insanely lucid dreams about Matthew McConaughey riding a llama to spontaneous phobias of ground beef, sagging skin and rapid muscle lossWeekly weight loss shots that have been hailed as 'miracle' injections are causing a host of nasty side effects, ranging from the unusual to the outright frightening.
More than five million Americans are taking blockbuster drugs like Wegovy and Ozempic to achieve svelte physiques since they started to be rolled out in 2021.
But there are now growing reports of patients suffering physical and psychological side effects from the brand-new drugs, with doctors still trying to decipher the cause.
A TikTok group has been set up to chronicle Ozempic dreams, with social media users detailing how they started experiencing insanely vivid dreams, including one user who watched Matthew McConaughey ride a llama and being rescued by Oprah Winfrey on a go-kart. Patients have even said the drugs have also become spontaneously repulsed by their favorite foods.
Users of the drug have also found they are suffering rapid muscle loss, tending to lose more muscle than fat while on the drug and bounce back weight gain when they come off it, with some gaining back more weight than they'd lost.
Wegovy and Ozempic are medications that work by tricking someone into thinking they are already full by mimicking hormones. This cuts someone's calorie intake, leading to them losing weight rapidly
Odd dreams reported have included seeing Oprah Winfrey ride a go-cart
Wildly vivib dreamsA number of patients have reported weird dreams while on the drug Wegovy, that tend to involve celebrities.
These include visiting a cattle auction with actor Matthew McConaughey to buy a llama, rob a museum with Jennifer Lopez or be rescued by a go-kart riding Oprah Winfrey.
Others mentioned by patients include telling Dwayne 'the Rock' Johnson that they are having his baby or even just constantly taking a peaceful drive on a sunny day.
There is even a TikTok handle dedicated to the dreams of Ozempic users — called OzempicDreams.
Experts are not clear exactly what's causing this, but say it may be linked to the impact the drug has on a patient's brain.
Dr Caroline Apovian, a medic at Harvard Medical School in Massachusetts, told the Wall Street Journal that it may also be caused by users expending more energy while they are asleep or finding it easier to recall dreams than before.
Repulsed by favorite foodsPatients on Wegovy say they are now starting to be disgusted by their favorite foods and some items that they never thought twice about.
Staci Rice, 40, from Georgia, lost nearly 50 pounds when she went onto Ozempic and can now fit into jeans she last wore 16 years ago.
But the marketing professional was also surprised to find that she had developed an aversion to ground beef and Chick-fil-A while on the drug.
Ground beef has now been pulled from dinners, must to the frustration of her husband and son, she told the Insider. And she is now also having Chick-fil-A's kale salad instead of its standard 'Number 1'.
She was also a lifelong coffee drinker, having enjoyed a cup every day since the seventh grade. But now, she can't touch it.
'Every morning, I would try to make coffee, thinking that one day it would just taste good to me again,' Ms Rice said.
Staci Rice, 40, (pictured) lost 50lbs after using Wegovy for six months. But she said she also became repulsed by coffee and other snacks that she used to enjoy
Kait Morris, a TikToker with more than 6,000 followers, has also revealed online how taking Ozempic led to her now being unable to eat meat and most other foods — and instead only craving smoothies.
She said: 'I miss eating, I miss going out to restaurants, I miss ordering a normal plate of food.'
Scientists suggest that patients may develop disgust for foods while on the drug because it can alter how the brain perceives something in the mouth. But the actual cause is not clear.
The condition causes certain foods and drinks to seem sweet, sour, bitter or metallic.
Saggy skinSagging skin has been reported as a side effect of the drug
Patients who are using weight loss drugs are facing saggy skin, doctors warn, which has been dubbed 'Ozempic face' and 'Ozempic body'.
It is caused by rapid weight loss that happens so quickly that the skin does not have time to adjust to the new body size. As a result, it hangs down in folds.
Dr Simon Ourian, a cosmetic dermatologist in Beverly Hills, California, revealed the effect, saying: 'If you lose a lot of weight quickly, your skin gets saggy, everything from the face to the buttocks is deflated.'
Patients affected by this include Nika Steward, 39, from North Carolina, who says she is 'not a fan of how I look' after a 100lb weight loss left her with excess skin.
Patients are planning fillers and tummy tucks to help get rid of the excess skin.
Doctors also say that patients using Wegovy and Ozempic may lose more muscle than fat
Rapid muscle lossWhen people see the ticker on the scale heading downward, they assume that all they are losing is fat.
But doctors have found that this is not the case — and that most patients are in fact losing more muscle than fat.
Dr Florence Comite, an endocrinologist in New York who has treated hundreds of patients, revealed to DailyMail.Com: 'What we see here at the center, it is usually more muscle loss than fat. At least 50 percent of our patients will lead with muscle loss.'
When people are eating a calorie-restricted diet, the body is more likely to remove muscle than fat because it is more expensive to maintain.
The patient will also likely be consuming less protein and not doing enough resistance training, which will also drive muscle loss.
Some doctors are now avoiding prescribing semaglutide to patients who are a healthy weight, raising concerns over the effects of muscle loss.
There are also concerns that muscle loss among older patients using the drugs could leave them struggling with balance or daily chores more than beforehand.
The weight loss drugs could cause thyroid cancer, doctors warn
Cancer warningThere are concerns that the weight loss drugs could trigger thyroid cancer in very rare cases.
The thyroid produces hormones that regulate heart rate, blood pressure, temperature and weight.
But scientists fear that when people take the medications this may 'overstimulate' the gland, leading to the development of tumors.
Doctors in the UK are investigating whether the drug can cause cancer and in the US it is included as a warning on the packaging. People with a family history of thyroid cancer are advised not to take the medication.
Scientists at NYU Langone Health say studies in mice have pointed to the link but that it is yet to be shown in human trials. They say the risk is currently being highlighted out of an 'abundance of caution'.
About 44,000 Americans are diagnosed with thyroid cancer every year, statistics suggest, while 2,100 die from the disease.
Metallic taste in mouthMany Wegovy users have also reported experiencing a strange metallic taste in their mouth while on the medication.
They include Jamie Walters, 35, from the US said she had the 'most gross metal taste ever' in her mouth one morning when she woke up.
Experts say the condition may be because the drug is affecting taste receptors in the mouth.
There is also a condition known as diabetic tongue, which can also lead to people having a metallic taste in their mouths.
This happens because of too much sugar in the saliva and a dry mouth, which can trigger oral thrush.
Fat bounce backMany patients have boasted on social media about losing tens of pounds while on Wegovy and Ozempic — and some more than 100.
But few have mentioned that once coming off the drug they piled a lot of the weight — if not more — back on.
A UK study found that people who used Wegovy experienced rapid weight loss, dropping 18% of their weight over 68 weeks. They regained two-thirds of that weight, or 12% of their original body weight in the year after dropping the weekly injections. Experts says the drug needs to be used over a lifetime to keep off the pounds
A study by the University of Liverpool, in the UK, published in April found that patients who used the 'game-changer' drug dropped 18 percent of their body weight in 68 weeks.
But, after dropping the weekly injections, users would put back on two-thirds of the weight within the next year. There were 336 people in the trials.
Among those to reveal they gained back weight after coming off the drug is model Remi Bader, who says that once she stopped using Ozempic she gained back 'double' what she lost while on the drug.
Experts say people who use the drug may pile the pounds back on after because it has not addressed the underlying issue causing them to over-eat.
They also say that muscle loss triggered by the drug means someone who uses it cannot return to their old diet because it will lead to weight gain.
Despite being hailed as one of the most powerful pharmaceutical tools to date, experts have warned it is not a 'magic pill' or miracle fix-all. Trials have shown that users can rapidly pile pounds back on once they stop taking the fat-fighting drug and it can trigger a variety of nasty side effects. Users commonly complain of nausea, constipation and diarrhea after taking the medication
Wegovy works by triggering the body to produce a hormone called glucagon-like peptide-1 that is released naturally from the intestines after meals
Red wine is rich in a chemical component called resveratrol which, in small doses, imitates oestrogen and triggers production of anti-ageing proteins called sirtuins. Sirtuins help protect against diseases such as Type 2 diabetes, osteoporosis, the metabolic syndrome, inflammatory, Alzheimer's and heart diseases (stock photo)
No alcohol cravingsIf you're someone who enjoys a nightly beer or glass of wine, then look away now!
There are hundreds of anecdotal reports on social media from patients saying Wegovy and Ozempic actually led to them avoiding alcohol.
One Ozempic user said on social media just this week: 'I haven't had a drink in 15 days! I used to drink every night and thought I could never stop.'
Eva Monsen, 46, also detailed to the New York Times how she 'almost immediately' kicked the habit of regularly drinking wine to help her relax while on Ozempic.
Experts are not certain why the drug may be putting people off alcohol but say that it could be linked to changes in areas of the brain that regulate the desire for food.
The Centers for Disease Control and Prevention (CDC) says women should not have more than one alcoholic beverage a day while men should not have more than two.
But a growing body of literature suggests even this is too much, with Canada now lowering its recommendation to no more than two drinks a week.
Llama Owner Blames Portland Police Dog For Livestock Death
Your browser does not support the audio element.
As temperatures in Warren, Oregon, hit 99 degrees on July 25, 2022, Portland Police Officer Christopher Verbout brought his narcotics dog Stitch inside his home to help keep the animal cool.
Normally, and in accordance with the Portland Police Bureau canine unit's standard operating procedures, Stitch was kept in a fenced-in kennel at Verbout's home 20 miles north of Portland in Columbia County. That day, Verbout chose to make an exception.
According to a written statement Verbout later gave to the Portland Police Bureau, when his wife came inside, Stitch "pushed by my wife through the open front door and ran out of the house."
Stitch is a Belgian Malinois, a breed of smart, loyal working dogs frequently used by police and military. Verbout told investigators he chased Stitch as she jumped a 6-foot-high fence, entered his neighbor Yvonne Pea's property and pursued Oreo, Pea's 14-year-old llama. Stitch chased Oreo over two more fences. The dog latched on multiple times, biting Oreo's legs and clamping her jaws around Oreo's neck, according to witnesses.
This February, extreme temperatures again hit Warren, this time plummeting to the low teens. Oreo died from hypothermia. After the attack last summer, Pea said, they had to shear Oreo outside of his normal schedule in order to assess his injuries. As a result, his wool did not grow back in time for winter. Pea said the thin coat, combined with his rapid weight loss in the weeks after the attack, led to his fatal hypothermia.
Oreo, a llama who was attacked by a Portland Police dog in July 2022, sits in his heated shelter with Miss Piggy Sue, a pig, in Warren, Ore. Oreo's owner said he had to be shorn after the attack to check for injuries and his thin wool coat led to his death February death from hypothermia.
Courtesy of Pea family
"I 100% blame you and the incident last July," Pea wrote to Verbout in an email days after Oreo died.
The July incident is at least the second time in two years a Portland police dog has escaped its handler's home and attacked a person or animal. The two attacks come as the use of police dogs nationwide is coming under closer scrutiny and a growing number of unintentional bites call into question how much control officers have over their dogs.
A Clackamas County Sheriff Deputy's canine bit a Portland Police officer in 2018, while the officer was helping arrest a man who had fled a stolen vehicle. It took the deputy about 30 seconds to get his dog to release the officer's leg, leaving a serious injury. An NPR investigation also found police dogs are frequently used against people who don't pose a threat and that the dogs are more violent and inflict more severe injuries than their human partners. NPR also found police dog handlers often, in private, express concern about getting their dogs to release on command. In 2016, 190 law enforcement officers in California reported being injured by their dogs.
And in Georgia, a sheriff's deputy was forced to shoot and kill his own dog when the animal bit the deputy's leg. The deputy said it was a case of mistaken identity.
Handling dogs that attackOregon state law affords some leeway for counties to decide what to do when a dog — any dog, not just one working for law enforcement — injures livestock.
For a dog's first offense, Columbia County is required to impose a fine between $250 and $1,000. Commissioners also have the option of requiring the dog be surrendered or removed to a location where it can't threaten livestock. If other options aren't available, county leaders can have the dog euthanized.
Although drug dogs aren't trained to bite, the Police Bureau has procedures in the event one does. A dog handler is required to photograph any injuries, provide treatment as necessary, report the bite to their supervisor and write a summary of what happened. A supervisor has to write a dog bite summary report which is reviewed by an assistant chief. The handler and the dog are evaluated by a trainer to determine if the dog is becoming more aggressive and if the handler can control their dog.
Police Bureau spokesperson Lt. Nathan Sheppard said the bureau anticipates a lawsuit in Stitch's case and declined to answer a list of questions about the incident.
Last July, as Verbout chased Stitch across Pea's property, she said Verbout looked like a "scared bunny." He "had absolutely no control of the dog physically or verbally," she said.
"When my husband, Terry, caught up to the animals, [Verbout] was tugging the dog, with his arms wrapped around the dog, trying to get it off the llama's side."
Columbia County Commissioners held a hearing to determine what penalties would be imposed against Stitch in August 2022. In written testimony, Pea's daughter, Jessica Pea-James, said she saw Stitch dangling from Oreo's neck and doing a "death shake." She watched Verbout struggle to free the dog from Oreo for about two and a half minutes.
"I never heard a release word or command, just Chris Verbout muttering over and over, 'please let go,'" she wrote.
In a written statement after the incident posted on the county website, Verbout said only that when he caught up to Stitch and Oreo, he put the dog back on her leash and returned her to his property.
"I never pried the jaws open," Verbout told commissioners. "I grabbed ahold of her and pulled her off…I didn't see any blood. I truly believe she was just attached to fur."
Testifying at the county hearing, Verbout described Stitch as "collar smart," explaining she behaves better when wearing her electronic collar that can shock her when she misbehaves. Stitch wasn't wearing her collar when she escaped Verbout's home.
A photo of Stitch, a Portland Police narcotics dog, provided to Columbia County commissioners in August 2022. Stitch escaped from her Portland Police officer handler's home and attacked the neighbor's llama in July 2022.
Columbia County Records
He told commissioners that during a later evaluation with her collar on, Stitch stopped mid-sprint while chasing a tennis ball and returned to Verbout on command. He said in the future, Stitch will have her collar on any time she's out of her kennel. That, he said, should ensure her obedience.
"I would recommend that she be able to behave without a collar," said County Commissioner Margaret Magruder, who was less forgiving of the dog's work life. "Dogs can do that."
Oregon state law requires any dog to be impounded if it kills, wounds, injures or chases livestock, llamas included. The county animal control officer, Roger Kadell, wrote in his report that he had Stitch impounded in Verbout's home. Kadell seemed to acknowledge it was an unusual step taken "with the special circumstances for this case as the dog involved was a working dog and taking the dog out of service for 30-45 days would not be in the public interest."
Kadell did not return messages asking to speak about this incident. County commissioners disagreed with the special treatment.
"The dog should have been immediately taken into custody because that's what we do with all dogs," then County Commissioner Henry Heimuller said during the hearing. "I don't think that we should create new processes on the fly for these kinds of things."
Stitch is obviously an aggressive dog, Heimuller said, describing her as a "highly trained, very expensive police officer for the City of Portland."
At the August hearing, Kadell, the animal control officer, said the dog had injured Oreo and that there's no guarantee it won't happen again.
"They're dogs, they can bite," he said.
Stitch's assessmentStitch is a drug detection dog trained to search cars, homes and people for narcotics and other contraband. The bureau also has patrol dogs trained to pursue and, at times, bite people. The bureau's dogs are owned by the city of Portland but live with their human partners and handlers, an arrangement typical for working dogs and one which deputy city attorney Michael Porter told county commissioners is critical to their relationship.
Stitch was purchased from the Springfield Police Department and partnered with Verbout in October 2021, about nine months before Oreo was attacked. She and Verbout trained together for two weeks before passing the Oregon Police Canine Association Detection Dog Certification Test the following month.
According to his statement, at the time of the hearing, Verbout and Stitch had over 149 hours of drug detection training together. The many certifications Verbout submitted as evidence of the dog's training all pertained to drug detection, not obedience.
Belgian Malinois occasionally have a gene which researchers at the University of California, Davis, found was associated with increased reports of "seizure, 'glazing over' behaviors, episodic biting behaviors, and general loss of clarity." The university's veterinary genetics laboratory sells a $50 test for the gene. The Portland Police Bureau didn't respond to questions asking if Stitch or any other Belgian Malinois at the bureau had been tested.
After the incident, Portland police Sgt. Jeffrey Dorn assessed Stitch at the request of Police Bureau Cmdr. Art Nakamura. Dorn is a sergeant with the bureau's Canine Unit. The review included three officers playing with Stitch in a conference room, an obedience assessment, and time interacting with other police dogs.
"We observed no unwanted behaviors from Stitch and Ofc. Verbout appeared to have acceptable control of her," Dorn wrote in his assessment, which was provided to Columbia County. "The Oregon Police Canine Association standards for narcotic detection dogs do not have a set of obedience standards that they train to so it was not possible to put her through the same tests that those on the patrol side train to."
Dorn's police dog has also attacked neighbors. After escaping his Happy Valley home in March 2021, Dorn's German Shepherd attacked and severely injured two people. According to a lawsuit filed last month against Dorn, his wife, and the City of Portland, the dog lunged at Kristina Norris, bit her arm and dragged her to the ground. Her husband Jason Norris was attempting to pull the dog off her when the dog let go and bit his lower leg and attempted to drag him.
"The dog mauled Mr. Norris for several minutes while both he and Mrs. Norris screamed for help and tried desperately to end the vicious attack," their lawsuit says.
The couple is seeking $1.7 million.
Both the incident with Stitch and the llama and the attack by Dorn's police dog appear to violate several canine unit standard operating procedures put in place by the commander of the Specialized Resources Division, the police bureau group that oversees canine officers. Those procedures include requirements that officers "exercise control over his or her canine at all times," and to always "keep the canine locked in its home kennel or a fence-controlled area when off duty."
Standard operating procedures are not the same as binding policies. Portland's Independent Police Review, the group that investigates police policy violations, told Pea that because no policy was violated, they could not recommend a formal investigation.
Sgt. Ryan Derry, who did the bureau's internal investigation into the llama attack, called the incident "an isolated event put into motion by a series of abnormal factors."
"The Verbouts are a family that takes great pride in Stitch as well as the other animals they have on their property," Derry wrote in the bureau's after-action review. "I can tell after speaking with them they feel terrible about what happened, not only for letting Stitch escape but also for the health of Oreo."
Columbia County commissioners voted 2-1 to allow Stitch to remain with Verbout on the condition that the city of Portland paid to install electric fencing around the property, Stitch and Verbout received obedience training, and the city implemented the animal control officer's recommendations. Those include ensuring the dog remains in her kennel or under the control of an adult at all times and that she is leashed when inside the house or going to and from a car. Portland Police Assistant Chief Jami Resch told commissioners the bureau leadership would be inspecting all canine kennels, and they were planning to review narcotics dog training "to make sure that they are receiving the highest level of training."
When asked this month, the bureau did not say if those reviews had taken place.
Magruder, who along with being a county commissioner is a lifelong farmer who has raised livestock, wanted Stitch removed and relocated someplace where the dog couldn't harm farm animals. She told OPB, in her experience, "dogs do the job they're trained to do, and if they don't, we don't keep them."
The county also imposed a $1,000 fine and ordered Portland to pay for all medical bills associated with the attack.
"I have a check that they sent me, and I just tucked it in my box," Pea said. "I didn't know if I signed it, if that meant I was closing the case."
She said she doesn't want to sue the city but wants Stitch removed from her neighbor's property.
RedPajama Replicates LLaMA Dataset To Build Open Source, State-of-the-art LLMs
Join top executives in San Francisco on July 11-12, to hear how leaders are integrating and optimizing AI investments for success. Learn More
Thought the open source AI references to camelids were finished? Think again: Yesterday, Together, a Menlo Park, California-based company focused on building a decentralized cloud and open source models, announced RedPajama (yes, like Llama Llama Red Pajama) yesterday.
"In many ways, AI is having its Linux moment," the company said in a blog post, linking to a January post written by Chris Re, co-founder of Together, Stanford associate professor and co-founder of SambaNova, Snorkel.Ai and Factory.
RedPajama is a collaborative project between Together, Ontocord.Ai, ETH DS3Lab, Stanford CRFM, Hazy Research, and MILA Québec AI Institute to create leading, fully open-source large language models (LLMs). Its effort began with yesterday's release of a 1.2 trillion token dataset that follows the LLaMA recipe. The data enables any organization to pre-train models that can be permissively licensed. The full dataset is available on Hugging Face and users can reproduce results with Apache 2.0 scripts available on Github.
LLaMA is a state-of-the-art foundation LLM released in February by Meta with gated access to researchers. Several other models based on LLaMA have come out in recent weeks, including Alpaca, Vicuna and Koala — but those models have not been available for commercial use. There was also some LLaMA-drama when the LLaMA model was leaked on 4chan.
EventTransform 2023
Join us in San Francisco on July 11-12, where top executives will share how they have integrated and optimized AI investments for success and avoided common pitfalls.
Register NowIn the coming weeks, Together will release a full suite of LLMs and instruction tuned versions based on the RedPajama dataset. The company emphasized that the forthcoming models will be fully open-source and commercially viable. In a tweet, the company said, "We hope this can be a clean-room, drama-free version. The RedPajama models we release, starting in the coming weeks, will be released under the Apache 2.0 license."
RedPajama part of a wave of open source AIAs VentureBeat reported last week, open source AI has been having a moment over the past few weeks, following the wave of LLM releases and an effort by startups, collectives and academics to push back on the shift in AI to closed, proprietary LLMs.
And a camelid-adjacent model, Dolly 2.0 (as in Dolly the Sheep), also made headlines last week when its developer, Databricks, called it the first open, instruction-following LLM for commercial use.
But the largest, state-of-the-art open source LLMs like LLaMA have been limited to the research community. "They are limited in that you can't build real applications and ship them," said Vipul Ved Prakash, founder and CEO of Together and previously cofounder of Cloudmark and Topsy. "We think having permissively licensed models is a critical aspect of open source AI."
Replicating the LLaMA dataset was no small taskThe company started with LLaMa, which it called the "leading suite of open base models," because it was trained on a "very large dataset that was carefully filtered for quality." Also, the 7 billion parameter LLaMA model is "trained for much longer, well beyond the Chinchilla-optimal point, to ensure the best quality at that model size."
While neither the dataset nor the model will be identical, the developers aim to create a fully open source reproduction of LLaMA which would be available for commercial applications, and provide a "more transparent pipeline for research."
The developers did not have access to the LLaMA dataset but had enough of a recipe to go on. "We followed the recipe very carefully to essentially recreate [the LLaMA dataset] from scratch," said Prakash. The dataset consists of seven data slices, including data from Common Crawl, arxiv, Github, Wikipedia and a corpus of open books.
"For each data slice, we conduct careful data pre-processing and filtering, and tune our quality filters to roughly match the number of tokens as reported by Meta AI in the LLaMA paper," read the blog post.
"All of the data LLaMA was trained on is openly available data, but the challenge was that they they didn't provide the actual data set — there's a lot of work to go from the overview to the actual data set," said Prakash. For example, he explained, the paper might describe how they picked the best 10,000 from a million documents, but they didn't give you the 10,000. "So we followed the recipe to repeat all that work to create an equivalent dataset," he said.
The debate over building transparent systemsPrakash said that the RedPajama project collaborators believe it's important that systems are transparent. "You know exactly how this model was built, what went into it," he said. "If you're trying to improve it, you can start from the dataset."
The project also brings together a larger community to these models, he added. "I would say academia has really been cut out of foundation model research because of the level of resources required, starting from data to the compute," he said. He added that there is a small number of people in the world working on these large models today, and if there was broader access, "a lot of brilliant people" around the world would be able to explore different directions of neural architectures, training algorithms and safety research.
"Also, this is one of the first really general AI which can be adapted to different tasks, and we think the applicability is very broad," he said. "But many different applications are possible only if you have access to the model, the model weights, and adapt them to different computing environments. We see a lot of this happen because of open source AI."
There is another side to the open source AI debate, however. For example, Ilya Sutskever, OpenAI's chief scientist and co-founder, recently said it was "wrong" to share research so openly, saying fear of competition and fears over safety — were "self-evident." He added that "at some point it will be quite easy, if one wanted, to cause a great deal of harm with those models."
And in a recent interview with VentureBeat, Joelle Pineau, VP of AI research at Meta, said that while accountability and transparency in AI models is essential, the key for Meta is to balance the level of access, which can vary depending on the potential harm of the model.
"My hope, and it's reflected in our strategy for data access, is to figure out how to allow transparency for verifiability audits of these models," she said, adding that access could be decided based on the level of potential harm of the model.
On the other hand, she said that some levels of openness go too far. "That's why the LLaMA model had a gated release," she explained. "Many people would have been very happy to go totally open. I don't think that's the responsible thing to do today."
Debates around ethical datasets as wellThere have also been debates about the ethics of the datasets themselves, whether the models are open or closed. An article last week in The Guardian said that the "enormous datasets used to train the latest generation of these AI systems, like those behind ChatGPT and Stable Diffusion, are likely to contain billions of images scraped from the internet, millions of pirated ebooks, the entire proceedings of 16 years of the European parliament and the whole of English-language Wikipedia."
But Prakash says that he thinks "these models capture in some ways the output of human society and there is a sort of obligation to make them open and usable by everyone." He added that "most of the magic" of these models comes from the fact that they are trained on "really broad and vast" data.
He also pointed out that the original data is compressed significantly in the actual model. The RedPajama dataset is 5 terabytes, and the models can be as small as 14 GB, ~500x smaller than the original data they are modeling.
"This means that knowledge from the data is abstracted, transformed and modeled in a very different representation of weights and biases of parameters in the neural network model, and not stored and used in its original form," said Prakash. So, it is "not reproducing the training data — it is derivative work on top of that. From our understanding, it is considered fair use as long as the model is not reproducing the data — it's learning from it."
There is no doubt that the open source AI debates are highly-complex. But when asked why the company called the new project RedPajama, the answer was far more simple. "A lot of us have small children," said Prakash. "It just seemed fun."
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.
Comments
Post a Comment