Farzin Zaker

VP of Engineering @ IgniteTech. Crafting AI-first solutions with a stellar team. Innovator, tech architect, AI enthusiast. Join us at

[Farzin Zaker @fzignite]

@0IvankaTrumpRV2 Absolutely

[Poonam Soni @CodeByPoonam]

Goodbye Photoshop

It's only been 2 days since Google dropped Gemini Flash Experimental, and people are going crazy over it.

13 wild examples so far (Don't miss the 5th one)

[Theo Jaffee @theojaffee]

I just resigned from @OpenAI. I was the guy responsible for communicating to the public that the voice feature that they have in the ChatGPT app isn’t the one they just demoed

[Farzin Zaker @fzignite]

The paper "Efficient CNN Design via Differentiable Neural Architecture Search" by #Meta explores new methods for creating efficient convolutional neural networks (#CNNs). The researchers utilize differentiable neural architecture search (#DNAS) to optimize architectures for both performance and resource efficiency. By integrating DNAS with other techniques, they aim to reduce computational costs while maintaining high accuracy. This approach could be impactful in real-world applications requiring a balance of power and performance.

#AIResearch #DeepLearning #NeuralNetworks #MachineLearning

Read more: https://t.co/ya6TuNnliw

[LlamaIndex 🦙 @llama_index]

Fully Local RAG 101 with Llama 3 + @ollama + @llama_index

This article by @pavan_mantha1 is a great handbook for creating a fully local RAG pipeline with @llama_index (with a HyDE layer).

It is lower-level than our “5 lines of code” Quickstart, giving you a better understanding of our lower-level components: text splitters, nodes/documents, @qdrant_engine integration, response synthesizer, and more.

https://t.co/EevL8az6VL

[Farzin Zaker @fzignite]

Another annoying error message from #ChatGPT! #OpenAI, Why should we lose access to our own #GPTs from time to time?

[Farzin Zaker @fzignite]

From auto-assigning tickets to offering insights on bug fixes and feature development, AI integration is reshaping our workflow like never before! Wondering how we ever managed without these recent GenAI advancements.
#AIRevolution #GenAI #IgniteTech #GFISoftware

[Artificial Guy / João Vitor A. @artificialguybr]

Gemini announced prices:

Gemini 1.5 Pro:
Free.
2 request per minute.
32k tokens per minute.
50 requests per day for free.

Pay as you go:
5 request per minute.
10M tokens per minute.
2k requests per day.
$7/1M Tokens INput.
$21/1M tokens output.

[Eric Vaughan @TheGenAICEO]

That’s a wrap at @ImagineAILive Incredible inaugural event. Honored to have been a part of the original, game changing event. #ai #imagineai @IgniteTech

[Eric Vaughan @TheGenAICEO]

At the opening reception at @ImagineAILive last night where @IgniteTech is a founding sponsor, we had amazing conversations with a cross section of people: LLM creators; an owner of a boutique financial services firm; a principal at a high school who is leading the charge to AI for his district. Quote: “ I know this is changing our lives, so I had to be here to learn what I need to do and how to drive this.” This is a special event because of the blend of experts, creators and business owners who have heard the siren call to act now.

[Farzin Zaker @fzignite]

Best #free #Google #AI courses:

#IgniteTech #GFISoftware

[Farzin Zaker @fzignite]

ScreenAI: the vision-language model transforming UI and infographics understanding! 🚀 With a unique training approach and at just 5B parameters, it's setting new benchmarks in the field. Dive into the future of human-machine interaction! #AI #ScreenAI #Innovation #IgniteTech #GFISoftware

Read More: https://t.co/9nwqnr89FK

[Farzin Zaker @fzignite]

Isn't it amazing? #Sora #OpenAI

#IgniteTech #GFISoftware

[Min Choi @minchoi]

3. Niagara Falls but with paint

Source: OpenAI TikTok

[Farzin Zaker @fzignite]

#OpenSora takes a giant leap for #AICommunity! They've gone FULL open-source with their video-generation model. Get your hands on model weights, training code, and architecture insights. Time to build our dream video-gen model! #AI #GenAI #OpenSource
#IgniteTech #GFISoftware

Repository: https://t.co/N2RbQPi1Jb

[Santiago @svpino]

AI will not replace you. A person using AI will.

[Farzin Zaker @fzignite]

Meet Devin, the AI revolutionizing software engineering! With top-notch skills in coding, debugging, and deploying apps, Devin sets new benchmarks in AI tech, outshining previous models. A leap towards future tech by Cognition Labs. #AI #SoftwareEngineering #TechInnovation #IgniteTech #GFISoftware

Read more: https://t.co/MBEfkShOvu.

[Farzin Zaker @fzignite]

#ElonMusk shakes the #AI world by making his chatbot #Grok #OpenSource, challenging #OpenAI's approach. With his startup xAI, Musk promotes transparency and innovation, reigniting the debate on AI's open-source ethics. #IgniteTech #GFISoftware

Read more: https://t.co/NnhOhue7nr

[Farzin Zaker @fzignite]

📷

What sets Zapier Central (Bots) apart from GPTs? Its array of diverse connectors simplifies the task of defining behaviors, offering ease even without a pre-existing set of actions. #Zapier #TechTrends #AIInnovation #IgniteTech #GFISoftware

More: https://t.co/elHQEKDJrC

[Mike Knoop @mikeknoop]

Zapier is opening a new era of AI Automation today with our public preview of Zapier Central -- a new workspace where you can build, teach, and work with AI bots that do useful work for you! Entirely through natural language, no code or tricky config required.

[Farzin Zaker @fzignite]

Anthropic introduces Claude 3 family: Haiku, Sonnet, & Opus models, redefining AI benchmarks with unparalleled intelligence, speed, and cost-efficiency across a wide range of tasks. Now available in 159 countries (Except #Canada), these models promise near-human comprehension, advanced analysis, and fewer refusals. #AI #GenAI #Claude3 #Innovations

Read more: https://t.co/fAGF4sWfKa

[Farzin Zaker @fzignite]

Traditional RAG is computationally far from optimal because the query and the documents must be encoded twice: once by the embedding model for retrieval and once by the language model as context. If we had the same model, we could cache (i.e., save for later) parts of the network during retrieval and re-use them during generation.

#Grit @RAG #GenAI #IgniteTech #GFISoftware

More: https://t.co/h72f7aLUQN