LAMA
product on 9 shows · 20 statements across 18 episodes · said 1 times in 1 episodes since 2024
WTF is with Nikhil Kamath 1 
the Y Combinator Startup Podcast 
Latent Space 
No Priors 
the MAD Podcast 
the a16z Podcast 
Big Technology 
All-In 
20VC
Mentions by year, every show
tap a year for its mentions
WTF is with Nikhil Kamath 1
every mention on every show, scene by scene, with the transcript →
20 statements about LAMA, every show
Stuart: GPU networking can consume up to 50% of LLaMA pre-fill runtime
“For example, networking can still consume up to 50% of total runtime for workloads like Lama's MDB pre-fill.”
Zeghidour: Meta's LLaMA and DINO models were developed in Paris
“Lama was started in Paris. Dino, which is the most groundbreaking vision work from Facebook was developed in Paris.”
Superhuman uses Baseten to run LLaMA and BERT classification models
“We use Base-Ten to run some I would say some LAMA, some BERT model for classification.”
Booz Allen deployed Meta's Llama to the ISS for zero-latency astronaut troubleshooting
“So we just put Lama on the international space station on the edge on satellites, right? So that, that enables the astronauts who are working on international space station to have Lama to chat with in space with no latency to determine when things go wrong, h…”
Morcos: Qwen is much easier to align than Llama due to pre-training
“It's much easier to RL Quen than it is to do Lama. Likely that has to do with the fact that Quen put a lot of synthetic reasoning traces into their training data.”
Krishnan: Passing SB 1047 Would Have Ended American Open-Source AI
“You know, like until last a year and a half ago, I was in California, along with all of you, California almost passed SB, which if that had happened, it would have been the end of open source, by the way, in the United States, we would not have a Lama. We woul…”
O'Driscoll: Meta won't build a major business selling LLaMA APIs
“It's not that they want to open source Lama and make money off it. They won't. It's not that they want to even have an API offering of Lama, something like, you know, a, an anthropic offering that just won't be big.”
Palicha: Zepto boosted its ad business using in-house Llama models
“We started training on Lama to try to build out a better, I mean, to basically build out the relevance engine in-house, and that gave us much better results, and we were able to surge the ads businesses out of that, so that had a big impact on bottom line”
LeCun: Enterprises prototype with proprietary APIs but deploy open-source models
“What we see is for, you know, partners who we talk to they say, well, our clients, when they prototype something, they may use a proprietary API, but when it comes time to actually deploy the product, they actually use Lama or open source, or other open source…”
LeCun: Meta's original LLaMA was built by 12 people in Paris
“Or another example of that is actually the first Lama came out of Paris. It came out of the FAIR labs in Paris. A small team of 12 people.”
Stebbings: Meta's productization of Llama AI models has been very poor
“I would say that Facebook is the same like Lama and the productization around Lama has been bluntly very, very poor at the moment.”
Appenzeller: Pre-training Meta's Llama models cost slightly over $3 million
“We ran the cost for some Lama models last year, and I think we ended up with, you know, a little over three million dollars, right?”
The top AI business model is fine-tuning open-source models for verticals
“The most likely business model that has to do with AI is taking a open source foundation model, like LAMA, which is the data open source system, which is used everywhere now, right? Every, almost every startup uses it even large companies. So take an open sour…”
Huang: NVIDIA Improved Hopper Performance on LLaMA 5x in One Year
“CUDA made it possible for us to iterate so quickly, just in the last year, and then we just went back and benchmarked when Lama first came out, we've improved the performance of hopper by a factor of five without the layer on top ever changing.”
Stebbings: Meta open-sources Llama because its true value is proprietary data
“Why does Facebook open source Lama? Because the values in the data that they have. A hundred percent.”
Gil: Cloud providers will capture substantial value by hosting diverse AI models
“And one could argue part of what that's going to do is just kind of flip some of the value capture, the revenue, the margin, the people, whatever metric you want to use over to the clouds, because they're going to be hosting all these things, right? So whether…”
Douwe Kiela: Meta's Original LLaMA Was Trained Entirely on Open Data
“So the LAMA model was not trained on any proprietary data. It was just trained on open data on the web.”
Douwe Kiela: Meta's LLaMA model used zero proprietary training data
“So the Lama model was not trained on any proprietary data. It was just trained on open data on the web”
Douwe Kiela: The open-source AI ecosystem exists solely due to Meta's LLaMA
“This whole flourishing that you see right now of open source models that basically comes from Meta's generosity in giving Lama away for free. And if they hadn't done that, then you wouldn't see that.”