Ashish Rathore

it's best not to describe urself and let the time do it for u.. 😎

[Ashish Rathore @aprathore09]

@HeyGen_Official The customer service is really bad. All my videos are getting rejected. And on top of that no one from customer support replies. They just ghost you once you make the payment.

[Ashish Rathore @aprathore09]

@HeyGen_Official really poor with the customer service. They have been rejecting my videos for no reason. Even one of them with random gibberish text was rejected because of moderation policy. Trying to reach out to these guys for 2 days but no response.

[Ashish Rathore @aprathore09]

Check out the demo video to see it in action : https://t.co/rKJzdnkiKV

Try it out for yourself here: https://t.co/umGkxRFwb9
(PS: Link only open for a day😅)

[Ashish Rathore @aprathore09]

🚀 I built a free AI-powered search engine using LangChain & GROQ, just like Perplexity! It offers accurate, contextual search results powered by advanced language models.

Link to demo video in comments👇.

#0to100xEngineer #GenAI #AI #OpenAIo1 @thesidboson @geeksiddhant

[Ashish Rathore @aprathore09]

Excited to share EmailGenie: AI-Powered Cold Email Generator!
Now with new features to make your outreach even more effective:
Versatile email types, Customizable inputs, Real-time follow-up interface

Check it out: https://t.co/NhLgKfaxT0

#AI #0to100xEngineer @geeksiddhant

[Ashish Rathore @aprathore09]

🚀 Introducing TeamGPT!
I built a free alternative to OpenAI's ChatGPT Team which costs $25/user/month. Now, you can harness the power of AI in your workspace—without paying a single penny! 💬
#0to100xEngineer #GenAI #chatgpt4 @100xengineers @geeksiddhant

[Ashish Rathore @aprathore09]

@MastersNitish @100xengineers @geeksiddhant Yeah that's true and that's why I kept only 3 as required and rest are optional

[Ashish Rathore @aprathore09]

Writing personalized emails used to be a headache for me . So, I built a GenAI tool that does the job for me! It handles tone, style, and everything in between, powered by Llama 3.1 & Mistral.
Check it out: https://t.co/FjK2hrHOW9
#0to100xEngineer @100xengineers @geeksiddhant

[Ashish Rathore @aprathore09]

Just tried FLUX.1 for image generation—great quality + accurate text. But then IDEOGRAM 2.0 launched last week, and it’s even better! 🤯

Check them out:
🔗 FLUX.1: [https://t.co/HU4iUFztUv]
🔗 IDEOGRAM 2.0: [https://t.co/qdLdgSNW6d]
#0to100xEngineer #flux1 #ideogram2 #AI

[Ashish Rathore @aprathore09]

Exploring Gradio to create UIs easily!

Gradio allows us to create UIs with just a few lines of code. I built a chatbot with Meta's Llama-3.1 to simplify complex financial terms.

More to come🙂 #0to100xEngineer
@100xengineers @geeksiddhant @thesidboson #Meta #llama31 #Finance

[Ashish Rathore @aprathore09]

Built a comfy workflow to transform myself into Lego characters!

Faced challenges making a Lego version of me, requiring separate LoRAs for Lego characters and my face. Checkout the results👇

@100xengineers #0to100xEngineer #GenAI #lora #ComfyUI

[Ashish Rathore @aprathore09]

Another difference is that Dreambooth fine-tunes the entire diffusion model, resulting in a model of the same size as the original (6.5GB in my case). In contrast, LoRA produces a much smaller fine-tuned model (around 55MB) but requires the base model to be used.

(3/3)

[Ashish Rathore @aprathore09]

Dreambooth Training: Never Seen Before Avatars of Captain Price (character from Call of Duty)

After a week of debugging, I fine-tuned the SDXL model to create Captain Price avatars using Dreambooth!

part - (1/3)
@100xengineers @thesidboson #0to100xEngineer #CallOfDuty #GenAI

[Ashish Rathore @aprathore09]

While both Dreambooth and LoRA allow for fine-tuning with a small dataset, Dreambooth requires regularization images (e.g., 200-400 AI-generated images of men if training on a male character). For my project, I used just 12 images of Captain Price.

(2/3)

[Ashish Rathore @aprathore09]

Getting started with ComfyUI.

ComfyUI is a simple yet powerful Stable Diffusion interface with a graph and nodes system. It allows you to connect models, prompts, and other nodes to create your own unique workflow.
#0to100xEngineer #ComfyUI #GenAI #stablediffusion @100xengineers

[Ashish Rathore @aprathore09]

Training a LoRA with SDXL checkpoint

LoRA lets us fine-tune large models like Stable Diffusion efficiently without needing a huge dataset. I trained a LoRA using just 14 images!

Check out the results👇 #0to100xEngineer #GenAI #LoRA #StableDiffusion @100xengineers

[Ashish Rathore @aprathore09]

Day 03 #0to100xEngineer challenge:
IP adapters was a struggle, but I finally got the results. In the future, we might not need models for dress photoshoots, AI can do it! And IP Adapters are just the start.
#GenAI #Claude @100xengineers #stablediffusion

[Ashish Rathore @aprathore09]

Day 3 of #0to100xChallenge: Discovering Inpainting! Altering images is even more exciting than creating from scratch. Inpainting lets you fix defects or add objects seamlessly. Built a product photoshoot engine with Stable Diffusion. Check the results👇

@100xengineers #GenAI #AI

[Ashish Rathore @aprathore09]

Day 02: Explored ControlNet and its variants. Retaining pose and structure from images was fascinating. Loved generating interior design ideas for my room, results were amazing! Check out these outputs. #0to100xEngineer #GenAI #StableDiffusion
@100xengineers

[Ashish Rathore @aprathore09]

Recently started my journey of exploring GenAI with 100Xengineers! Over the past few weeks, I've been learning about the diffusion model and getting familiar with Automatic1111.

Excited for what lies ahead!
#0to100xEngineer
@100xengineers