The model Databricks has is a streaming pile of shit that Naveen Rao sold you. It's worse than GPT-4 and Claude LLM models. GPT-4 is literally the best wrapper out there and unless you integrate it with Databricks there's no Gen AI story there. Databricks and Ali is all hype and it's going to fad when revenue numbers don't grow at the same pace that they did. Every company peaks out Databricks being in the B2B segment is going to peak out soon. 50% revenue growth is going to come down to 20% by the end of 2024-25. No one's buying your shit. Your Sales people are struggling to sell your Snake Oil Product. Databricks is literally Cloudera in the making. Their employees will come out here in droves to call me a bitter rejected candidate or some dude who the hiring committee downleveled. Like seriously I laugh when their employees don't even bother arguing on facts and instead just take personal potshots. No, I am not jealous of any of Databricks employees neither am I bitter about being rejected. I am just pissed to see Ali and all their Directors show up on my linkedin feed and media channels as if they are pioneering the AI wave with Naveen's garbage LLM Model.
lol you lost all credibility with “bitter about being rejected.” Why did you even apply if you don’t like the products?
Neither did I apply nor was I poached. I am just unhappy to see the leaders at the company fooling Google/AWS/Snowflake/Uber/Meta talent with their snakeoil product. That in my opinion is blasphemy.
OP smart ones never believed it in first place. Naive ones wont even listen to ur alarms
TC?
Didn’t gpt4 cost 100s of millioks to train? Databricks only spent like 15 million. They knew it wasn’t going to be better.
Think of DBRX as “proof in the pudding” any company using Databricks can train and serve a world class model on the platform using their first party data. After all we don’t care if you use DBRX or Mixtral or LLama/any other foundation model or an external model like GPT4 or Claude. You get a single place to train, serve, monitor, and fine tune YOUR model, which is really cool.
From all of the Media Galore, I only came to understand you guys are promoting DBRX as the foundational model by claiming how it's superior than the other ones available. And that's a lie.
Well, someone finally said "snake oil" I was amazed to hear a customer story on how much money they charge for their notebook experience. The same customer moved to GCP and reduced the cost by 40%.. Tbh.. thier long term growth is not sustainable, and Execs knows it very well, hence trying to pivot into LLM world to get more free investor money..
Thank God someone understands how these pre-IPO unicorns grift on VC Money creating hype and cashing out big during IPO. Like the RSUs they sell their employees are restricted units which you can't sell until 6 month lockup expiry period and that's when the price crashes causing the RSU holders to sell it a loss. Very few make money in an IPO. Ali is a snake-oil product salesman who somehow has managed to con investors by leveraging the Snowflake story.
Can you share deets about the notebooks experience? Dbx cost has always been opaque esp as it’s a cloud-prem offering
It's a great model with gpt3.5 quality(which is good enough for a lot of enterprise use cases when augmented with private data, fine tuning and RAG) and very efficient. Sorry that the company you hate is succeeding.
> It's a great model with gpt3.5 quality No it's not. It's performing much worse than GPT-3.5. Talk to customers using their model and then you will understand.
OP, what GenAI experience do you have ? How did you conclude the models are terrible ?
He is the CEO of generative AI
The same Uber guy who has been hating databricks on a lot similar posts again?
I haven’t used the model but anyone who thinks they can beat openai or anthropic or google are delusional. No one has an ai strategy. It’s mostly hype. This includes snowflake also btw.
Ofc atleast Snowflake is honest and stock is public for people to understand it's true value. What I get sad about is that Databricks employees, Directors and Recruiters hide behind a veil with fabricated numbers and fool hardworking talent from Meta/Google/AWS and trap them. That's an act of evil imo.
Well it works until it doesn’t. End of day the product speaks for itself and people will choose what’s better
ah it’s been a while since a good ol external databricks shitpost thanks op
Are you alright?
Calling out your lie. Argue on facts. Databricks is Cloudera in the making.
hard to debate such a solid fact as “u r cloudera”