The more I use Llama 3 the more I think that Zuck may have just killed OpenAI and all other large proprietary AI vendors. The gap between latest GPT4 and Llama 70b is virtually non existent. Even if OpenAI releases GPT5 now, 400b Llama 3 is still training and will most likely be…
179
521
4K
Replies
@mnkyho
you can achieve close to 300 tokens per second using
@GroqInc
which feels almost instant even for large amounts of text
1
0
42
@Lauramaywendel
Loads of fun seeing the competition!
Some cloud providers are providing abstraction layers to enable "cheap" LLM switching. Sounds like a win for devs.
I read AWS Bedrock has an API layer drawing out commonalities among their hosted LLMs. Makes our app integration layer…
2
0
28
@KenTabor
I agree reliability of consistent output is still an issue / sometimes you get unexpected behaviour. I think probably the only long term solution is to run a very cheap and fast model (e.g. Llama 8b) to do input / output validation / sanitization (esp for structured / json…
1
3
20
@Lauramaywendel
OpenAI turned against their developers once they went full B2C with ChatGPT and Dall-E, so anyway, although there's more competition, they will likely hold it with their B2C strategy. Microsoft went full B2B, but that makes sense because no matter what, at least in the short…
1
0
8
@Lauramaywendel
I agree - OpenAI needs to do something fast - they will start to see their developer community to start to migrate over to Open Source - if not careful
3
1
9
@Lauramaywendel
We’re in heaven with the Llama 3 release. Unlocks so many use cases for our clients. AI adoption is the slow roll part of AI deployment and open source makes it so much easier for us to implement across all security, privacy, and deployment use cases.
AI is not the product, at…
0
1
6
@Lauramaywendel
Great points, I agree 100%. I hope Llama 3 400b is also open sourced the same way. Very bullish on AI and Meta if they continue to open source these amazing models.
0
0
2
@Lauramaywendel
Google has unbelievable level of data organised, and habassis, so I’m not writing it off yet
1
0
4
@Lauramaywendel
You are missing one factor here, Microsoft backing OpenAI. Which has really deep pockets.
1
0
4
@Lauramaywendel
Engagement farming at its best and tell me you have no idea about the topic without telling me you have no idea about the topic.
1
0
1
@Lauramaywendel
Ofc GPT-5 will not be in the same ballpark with llama…it will be way more better
2
0
11
@Lauramaywendel
Llama 3 is a significant achievement in AI world. To know the actual current state of AI, refer 👇
Stanford 2024 AI Index Report Is Out -The Most Trustable Report
Here are the top 10 takeaways from the Stanford 500-page report on the current AI trends to know where to hit the hammer 🔨: 🧵👇
#GenerativeAI
#AI
#SoftwareEngineer
#ProductivityBoost
1
5
22
0
0
3
@Lauramaywendel
Sam Altman says GPT-4 kinda sucks and the model is over a year old. OpenAI still has a first mover advantage in my opinion. More competition will benefit all the AI vendors.
0
0
4
@Lauramaywendel
Mark really is the MVP in this AI era 😱
After open-sourcing Llama 3 I must say that I’m starting to think Mark Zuckerberg is a cool guy, ngl
12
36
140
0
0
4
@Lauramaywendel
Meta is now on the lead, the use case and adoptions can easily outpace any other models. The open-source strategy makes more sense when you look long-term.
0
0
5
@Lauramaywendel
👀 so LLMs become commodities and value migrates to applications and/or infrastructure.. sth sth $7 trillion semiconductor project..
0
0
3