Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

ml-sharp 🤝 FLUX.2 klein Inspired by 大雄, I’ve trained a LoRA for klein 9b that can repair 3D views from ml-sharp, fixing geometry issues and restoring original details. Weights: I will share the demo + inference code soon!

Cyril Diagne

25,177 subscribers

31,632 Aufrufe • vor 7 Monaten •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

It's out, and it comes with a demo🔥 Introducing Flux.2 [klein] Move ✨ > specialized move & resize LoRA for Flux.2 [klein] 9B > draw boxes manually or pick from auto-detected objects > works with fast 4-8 step inference 💨

It's out, and it comes with a demo🔥 Introducing Flux.2 [klein] Move ✨ > specialized move & resize LoRA for Flux.2 [klein] 9B > draw boxes manually or pick from auto-detected objects > works with fast 4-8 step inference 💨

Linoy Tsaban

59,059 Aufrufe • vor 5 Monaten

I Built an open source virtual try-on professional LoRA at fal Lora for Flux Klein 9b Edit, give it a photo of someone + 2 clothing items and it dresses them up Works with men and women, all kinds of outfits Weights are open on HuggingFace, and there's a playground you can try right now

I Built an open source virtual try-on professional LoRA at fal Lora for Flux Klein 9b Edit, give it a photo of someone + 2 clothing items and it dresses them up Works with men and women, all kinds of outfits Weights are open on HuggingFace, and there's a playground you can try right now

Lovis Odin

15,494 Aufrufe • vor 6 Monaten

Found a hardcore LoRA workflow! Generate a PLY point cloud --> Adjust angle in editor -->Refine with Qwen-Image-Edit-2511 Gaussian Splash LoRA. It accurately restores complex perspective shifts. As shown in the demo, it handles 3D rotation and can even restore high-def details from blurry close-ups. Within a 45° range, the consistency is unmatched. Huge thanks to 大雄 for this contribution! 🫡

Found a hardcore LoRA workflow! Generate a PLY point cloud --> Adjust angle in editor -->Refine with Qwen-Image-Edit-2511 Gaussian Splash LoRA. It accurately restores complex perspective shifts. As shown in the demo, it handles 3D rotation and can even restore high-def details from blurry close-ups. Within a 45° range, the consistency is unmatched. Huge thanks to 大雄 for this contribution! 🫡

Tongyi Lab

1,493,527 Aufrufe • vor 7 Monaten

Finally, Apple's ML-Sharp will run directly in the upcoming Python scripting module in LichtFeld Studio as a plugin. This is not another ML-Sharp UI! The LichtFeld release 0.5 will transform LichtFeld Studio from mere splat software into a comprehensive platform. I'm developing a plugin system that will require uploading Python plugin scripts to GitHub with all dependencies specified. In LFS, you'll simply need to type the GitHub repo URL, and it will import the extension. Dependencies are managed by the integrated UV packet management. You'll be able to run Colmap, SAM3, Depth Models, and more directly in LFS, and even create your own UIs. This change will be transformative for the 3DGS software ecosystem. The custom LichtFeld tensor library is already fully interoperable with PyTorch via dlpack, bringing all the ML power into LFS if desired. I'm extremely hyped and excited about these upcoming features.

Finally, Apple's ML-Sharp will run directly in the upcoming Python scripting module in LichtFeld Studio as a plugin. This is not another ML-Sharp UI! The LichtFeld release 0.5 will transform LichtFeld Studio from mere splat software into a comprehensive platform. I'm developing a plugin system that will require uploading Python plugin scripts to GitHub with all dependencies specified. In LFS, you'll simply need to type the GitHub repo URL, and it will import the extension. Dependencies are managed by the integrated UV packet management. You'll be able to run Colmap, SAM3, Depth Models, and more directly in LFS, and even create your own UIs. This change will be transformative for the 3DGS software ecosystem. The custom LichtFeld tensor library is already fully interoperable with PyTorch via dlpack, bringing all the ML power into LFS if desired. I'm extremely hyped and excited about these upcoming features.

MrNeRF

10,404 Aufrufe • vor 8 Monaten

Super clean and efficient meshes by an AI? YES! The typical 3D Generative AI solutions produce lots of artifacts and usually way to many polygons due to volumetric approaches. In comparison “MeshGPT creates triangle meshes by autoregressively sampling from a transformer model that has been trained to produce tokens from a learned geometric vocabulary. These tokens can then be decoded into the faces of a triangle mesh. This method generates clean, coherent, and compact meshes, characterized by sharp edges and high fidelity.” Surely it is limited by the trained vocabulary but various versions can be trained for specific sets to create generative model libraries for certain object groups. Very promising approach with the high quality.

Super clean and efficient meshes by an AI? YES! The typical 3D Generative AI solutions produce lots of artifacts and usually way to many polygons due to volumetric approaches. In comparison “MeshGPT creates triangle meshes by autoregressively sampling from a transformer model that has been trained to produce tokens from a learned geometric vocabulary. These tokens can then be decoded into the faces of a triangle mesh. This method generates clean, coherent, and compact meshes, characterized by sharp edges and high fidelity.” Surely it is limited by the trained vocabulary but various versions can be trained for specific sets to create generative model libraries for certain object groups. Very promising approach with the high quality.

René Schulte

20,772 Aufrufe • vor 2 Jahren

Google Veo3 creates beautiful base videos, but what if that’s not enough? I built a ComfyUI workflow that takes it further: 🏗 New structure with Flux (LoRA arch) 📦 Turned into 3D with Hunyuan3d 2 🔁 Integrated + relight via Flux, Controlnet, Denoise and Redux 🎞️ Finalized the video using Wan2.1 + CausVid + VACE The result? Custom, controllable, cinematic videos far beyond the original VEO3. ⚠️ There are still a few scale and quality issues I'm currently working on, but the core process is solid. 📹 I’ll drop a full video tutorial next week. 📁 In the meantime, you can download the workflows, links in the comments (I am using a H100 for it, but probably a A100 is enough).

Google Veo3 creates beautiful base videos, but what if that’s not enough? I built a ComfyUI workflow that takes it further: 🏗 New structure with Flux (LoRA arch) 📦 Turned into 3D with Hunyuan3d 2 🔁 Integrated + relight via Flux, Controlnet, Denoise and Redux 🎞️ Finalized the video using Wan2.1 + CausVid + VACE The result? Custom, controllable, cinematic videos far beyond the original VEO3. ⚠️ There are still a few scale and quality issues I'm currently working on, but the core process is solid. 📹 I’ll drop a full video tutorial next week. 📁 In the meantime, you can download the workflows, links in the comments (I am using a H100 for it, but probably a A100 is enough).

Lovis Odin

89,083 Aufrufe • vor 1 Jahr

Holy shit… someone just made machine learning click. Not static diagrams. Not math-heavy PDFs. Not black-box training. Real algorithms — training step-by-step — visually. It’s called Machine Learning Visualized and it lets you watch models learn in real time. Here’s why this is different: Instead of dumping theory first, it shows optimization happening live: • gradients moving • weights updating • decision boundaries shifting • loss decreasing • models converging You literally see learning happen. Everything is built from first principles: • Gradient Descent • Logistic Regression • Perceptron • PCA • K-Means • Neural Networks • Backpropagation No magic. Just math → code → visualization. Each chapter is a Jupyter notebook that derives the math then implements it then animates training. So you can watch: • neural nets shape decision surfaces • PCA rotate feature space • K-means clusters form live • gradient descent find minima • sigmoid reshape boundaries • backprop update weights step-by-step This solves a huge problem: Most ML resources teach: math → code → ??? → trained model This shows: math → code → learning process → result Which means you finally understand: • why gradients matter • how weights evolve • what loss landscapes look like • how convergence actually happens • why deep nets learn non-linear functions Even better: You can open any notebook modify parameters and watch behavior change instantly. Learning ML becomes interactive. Not passive. Not abstract. Not confusing. Just… visible. Perfect for: • beginners learning ML • devs moving into AI • interview prep • teaching concepts • understanding backprop • visual learners • building intuition This is the kind of resource that makes neural networks finally “click”. Link: We’re moving from: reading about ML → watching ML learn That’s a big shift. Because once you can see training, you stop memorizing… and start understanding. AI education just got visual.

Holy shit… someone just made machine learning click. Not static diagrams. Not math-heavy PDFs. Not black-box training. Real algorithms — training step-by-step — visually. It’s called Machine Learning Visualized and it lets you watch models learn in real time. Here’s why this is different: Instead of dumping theory first, it shows optimization happening live: • gradients moving • weights updating • decision boundaries shifting • loss decreasing • models converging You literally see learning happen. Everything is built from first principles: • Gradient Descent • Logistic Regression • Perceptron • PCA • K-Means • Neural Networks • Backpropagation No magic. Just math → code → visualization. Each chapter is a Jupyter notebook that derives the math then implements it then animates training. So you can watch: • neural nets shape decision surfaces • PCA rotate feature space • K-means clusters form live • gradient descent find minima • sigmoid reshape boundaries • backprop update weights step-by-step This solves a huge problem: Most ML resources teach: math → code → ??? → trained model This shows: math → code → learning process → result Which means you finally understand: • why gradients matter • how weights evolve • what loss landscapes look like • how convergence actually happens • why deep nets learn non-linear functions Even better: You can open any notebook modify parameters and watch behavior change instantly. Learning ML becomes interactive. Not passive. Not abstract. Not confusing. Just… visible. Perfect for: • beginners learning ML • devs moving into AI • interview prep • teaching concepts • understanding backprop • visual learners • building intuition This is the kind of resource that makes neural networks finally “click”. Link: We’re moving from: reading about ML → watching ML learn That’s a big shift. Because once you can see training, you stop memorizing… and start understanding. AI education just got visual.

Suryansh Tiwari

132,607 Aufrufe • vor 5 Monaten

Elon, to be frank- your platform hates original creators. You shared my content that someone else stole without crediting. Now they will see the fruit of my labor. For full transparency, I received $900 for the past two weeks on this platform from receiving 100 million views. For reference, I’ve usually only seen $150 payout from X. On TikTok, I am at $1600 for the month with 750k views. Crazy difference. I’ve been putting in the work and a share from you could be life changing for me. Yet, you shared content from an already big account that put zero effort to add some emojis. I’m just so disappointed in the treatment of small accounts.

Elon, to be frank- your platform hates original creators. You shared my content that someone else stole without crediting. Now they will see the fruit of my labor. For full transparency, I received $900 for the past two weeks on this platform from receiving 100 million views. For reference, I’ve usually only seen $150 payout from X. On TikTok, I am at $1600 for the month with 750k views. Crazy difference. I’ve been putting in the work and a share from you could be life changing for me. Yet, you shared content from an already big account that put zero effort to add some emojis. I’m just so disappointed in the treatment of small accounts.

Anna Matson

177,873 Aufrufe • vor 1 Jahr

we achieved some progress in the implementation of state inference inside octra , first we managed to implement continuous reasoning without clearing the circle memory, second we completed a pure resource model under which validators can verify and certify the execution flow of a set of useful work (in this case this is ML inference with text generation by input tokens fully onchain), a few months ago the answers of the test model were one-word and substantially limited, in the current version (which will be available for everyone) the reasoning flow is automatically extended, it will be possible to upload various models into circles or create tools for their use making them public or only private (full demo will be opensource as soon as we finish, it will be possible to deploy anything in 2-3 clicks) live use case, the economic model also converges here - wallet pays for inference in native coins, validators (with GPU support can maintain availability of their resources to quickly process such private calls, while the execution core remains private through HFHE)

we achieved some progress in the implementation of state inference inside octra , first we managed to implement continuous reasoning without clearing the circle memory, second we completed a pure resource model under which validators can verify and certify the execution flow of a set of useful work (in this case this is ML inference with text generation by input tokens fully onchain), a few months ago the answers of the test model were one-word and substantially limited, in the current version (which will be available for everyone) the reasoning flow is automatically extended, it will be possible to upload various models into circles or create tools for their use making them public or only private (full demo will be opensource as soon as we finish, it will be possible to deploy anything in 2-3 clicks) live use case, the economic model also converges here - wallet pays for inference in native coins, validators (with GPU support can maintain availability of their resources to quickly process such private calls, while the execution core remains private through HFHE)

λ

11,482 Aufrufe • vor 1 Monat

People are still sleeping on WebMCP.. You can now give agents tool calls to interact with your website natively. Including full UI/UX support and interactive elements to build their own views. Sharing a demo below that I submitted for the WebMCP challenge - see how agents can interact with the El Nino situation tracker. It creates it's own views, preserves human edits, and creates a shareable link to share with other agents (or humans). Video and code by Codex ChatGPT & Railway Live at elneenyo dot com to try out today!

People are still sleeping on WebMCP.. You can now give agents tool calls to interact with your website natively. Including full UI/UX support and interactive elements to build their own views. Sharing a demo below that I submitted for the WebMCP challenge - see how agents can interact with the El Nino situation tracker. It creates it's own views, preserves human edits, and creates a shareable link to share with other agents (or humans). Video and code by Codex ChatGPT & Railway Live at elneenyo dot com to try out today!

Nikunj Kothari

20,897 Aufrufe • vor 7 Tagen

, webgl experience fully made using fal the pipeline: GPT-Image-2 (or FLUX.2 Klein 9B for the live "create your own") 🦴 meshy v6 image-to-3d takes that → rigged GLB with a baked animation 🐾 GPT-Image-2/edit on the same character → companion creature image, color-locked to the character's palette → meshy v6 again (no rig) → static creature GLB 🪨 PATINA (fal-ai/patina/material/extract) takes the character image and outputs basecolor + normal + roughness + metalness for a per-character floor that matches their environment 🎬 Seedance 2.0 fast → 8s looping background video per character 📦 everything runs through gltf-transform (resize 1024 → webp q80 → draco) for ~95% size reduction → 400-800 KB GLBs 🖥️ the front-end is a single static index.html: Three.js scene, alpha-mapped marble disc with PlaneGeometry tessellation for real displacementMap depth, mirrored reflection plane, breathing animations on companions, palette-swap via CSS custom properties on character click 🧩 the whole recipe is published as a Claude Code skill fal-regenerate-3d so you can rebuild this yourself → ⚠ metrics shown are decorative

, webgl experience fully made using fal the pipeline: GPT-Image-2 (or FLUX.2 Klein 9B for the live "create your own") 🦴 meshy v6 image-to-3d takes that → rigged GLB with a baked animation 🐾 GPT-Image-2/edit on the same character → companion creature image, color-locked to the character's palette → meshy v6 again (no rig) → static creature GLB 🪨 PATINA (fal-ai/patina/material/extract) takes the character image and outputs basecolor + normal + roughness + metalness for a per-character floor that matches their environment 🎬 Seedance 2.0 fast → 8s looping background video per character 📦 everything runs through gltf-transform (resize 1024 → webp q80 → draco) for ~95% size reduction → 400-800 KB GLBs 🖥️ the front-end is a single static index.html: Three.js scene, alpha-mapped marble disc with PlaneGeometry tessellation for real displacementMap depth, mirrored reflection plane, breathing animations on companions, palette-swap via CSS custom properties on character click 🧩 the whole recipe is published as a Claude Code skill fal-regenerate-3d so you can rebuild this yourself → ⚠ metrics shown are decorative

Lovis Odin

99,185 Aufrufe • vor 4 Monaten

I created a demo of a bioautomation system that uses LLMs, Opentrons, and lua to create a dynamic programming environment for cloud labs or robot/human clusters. It can reason about its own code based off of lab measurements. Most importantly, I actually fucking implemented it, and it is open source. Took about 4 days for this rough draft, and it is very much a draft. The user inputs their task, the system creates code, and then executes it. The code defines control flow from data generated in the lab. Not only can it create code, but it can reason about things that could have gone wrong, run analysis using an internal sandbox, and then create new code based off of that analysis for execution. Timestamps: 0:00 - intro and code generation 3:04 - homebrewed replacement for Opentrons API for running all this code 5:26 - dynamic control flow using data 6:10 - LLM reasoning about a biological protocol and fixing it 10:15 - rant on the future of cloud labs and bioautomation I made this as a demo for how I think we should be thinking about building and scaling biology. I believe we can encode the tacit knowledge of a laboratory into the knowledge of an LLM, that we can do reinforcement learning off of results it creates, and that we must do that by leveraging a sufficient quantity of unique, useful, verifiable protocols. That doesn't come from just doing drug screens - it comes from doing basic everyday experiments and doing them well. Through the elimination of tacit knowledge necessary to physically operate a lab + proper batching + models writing code, I think we can make building biotechnology 10x-100x cheaper and easier than it is nowadays.

I created a demo of a bioautomation system that uses LLMs, Opentrons, and lua to create a dynamic programming environment for cloud labs or robot/human clusters. It can reason about its own code based off of lab measurements. Most importantly, I actually fucking implemented it, and it is open source. Took about 4 days for this rough draft, and it is very much a draft. The user inputs their task, the system creates code, and then executes it. The code defines control flow from data generated in the lab. Not only can it create code, but it can reason about things that could have gone wrong, run analysis using an internal sandbox, and then create new code based off of that analysis for execution. Timestamps: 0:00 - intro and code generation 3:04 - homebrewed replacement for Opentrons API for running all this code 5:26 - dynamic control flow using data 6:10 - LLM reasoning about a biological protocol and fixing it 10:15 - rant on the future of cloud labs and bioautomation I made this as a demo for how I think we should be thinking about building and scaling biology. I believe we can encode the tacit knowledge of a laboratory into the knowledge of an LLM, that we can do reinforcement learning off of results it creates, and that we must do that by leveraging a sufficient quantity of unique, useful, verifiable protocols. That doesn't come from just doing drug screens - it comes from doing basic everyday experiments and doing them well. Through the elimination of tacit knowledge necessary to physically operate a lab + proper batching + models writing code, I think we can make building biotechnology 10x-100x cheaper and easier than it is nowadays.

Keoni Gandall

21,412 Aufrufe • vor 1 Jahr

WOW. 😳 Apple just quietly won the 3D maps war at WWDC. Gaussian Splatting is coming to Apple Maps Flyover this fall. Apple Maps Flyover covers 300+ cities. Until yesterday, every single one was built on standard drone photogrammetry. The technology captures photos from the air and reconstructs 3D geometry from them. Gaussian Splatting does not reconstruct geometry. It represents the scene as millions of tiny 3D ellipsoids, each one carrying its own color and opacity information based on how light actually behaves in that location. The output is not a mesh model. It is a field of light. When you move through it, it does not crumble at the edges. The detail holds because it was never geometry to begin with. Apple has been hiring for this for years. Their SHARP model, published in research last year, generates photorealistic 3D scenes from a single image in under a second. Google has more sensor data than anyone. More Street View cars, more satellites, more capture history. On navigation accuracy and geodata depth, Google Maps is still ahead by most measures. But fidelity in 3D city rendering is a different competition, and Apple just set a bar in that. Most people will experience this in the fall without knowing the name of the technology. They will open Flyover, look at a city they know, and notice it looks different. Real, not rendered. That is the moment Gaussian Splatting stops being a research term and becomes something a billion people use. Bookmark this. It will look prescient by October.

WOW. 😳 Apple just quietly won the 3D maps war at WWDC. Gaussian Splatting is coming to Apple Maps Flyover this fall. Apple Maps Flyover covers 300+ cities. Until yesterday, every single one was built on standard drone photogrammetry. The technology captures photos from the air and reconstructs 3D geometry from them. Gaussian Splatting does not reconstruct geometry. It represents the scene as millions of tiny 3D ellipsoids, each one carrying its own color and opacity information based on how light actually behaves in that location. The output is not a mesh model. It is a field of light. When you move through it, it does not crumble at the edges. The detail holds because it was never geometry to begin with. Apple has been hiring for this for years. Their SHARP model, published in research last year, generates photorealistic 3D scenes from a single image in under a second. Google has more sensor data than anyone. More Street View cars, more satellites, more capture history. On navigation accuracy and geodata depth, Google Maps is still ahead by most measures. But fidelity in 3D city rendering is a different competition, and Apple just set a bar in that. Most people will experience this in the fall without knowing the name of the technology. They will open Flyover, look at a city they know, and notice it looks different. Real, not rendered. That is the moment Gaussian Splatting stops being a research term and becomes something a billion people use. Bookmark this. It will look prescient by October.

Shruti

19,832 Aufrufe • vor 3 Monaten

A 36 years old named Mohammad Tariq Ali Quadri (Baba Khan) brutally murdered by a group of unknowns in #RamchandraNagar under #ISSadan police Limits, #Hyderabad, #Telangana. Tariq tries hard to save himself but a large number of miscreants surround him and start attacking him with sharp weapons. All the men can be seen fleeing from the spot on bikes after he died. The reason of murder is not yet clear and search is on for the miscreants who escaped after easily committing the crime. Police say that the accused will be arrested soon.

A 36 years old named Mohammad Tariq Ali Quadri (Baba Khan) brutally murdered by a group of unknowns in #RamchandraNagar under #ISSadan police Limits, #Hyderabad, #Telangana. Tariq tries hard to save himself but a large number of miscreants surround him and start attacking him with sharp weapons. All the men can be seen fleeing from the spot on bikes after he died. The reason of murder is not yet clear and search is on for the miscreants who escaped after easily committing the crime. Police say that the accused will be arrested soon.

Hate Detector 🔍

20,226 Aufrufe • vor 2 Jahren

As I promised yesterday, I'll briefly explain LoRA training and share a workflow I made so you can do it quickly. First, let me answer a very common question: 'Why train LoRAs when we have such advanced models?' Even though we have incredibly advanced models now (like NBP), we still can't always get them to do specific things we want. Simplest example: the spritesheet LoRA I made the other day. I generated 1000 images with Nano Banana and only 100 were what I wanted. The LoRA I trained using those 100 images gives me nearly 100% consistent results. Second point is cost and speed. With LoRA, we can cut costs by 4-5x. And while doing that, we're generating 4-5x faster. How many images do you need for a good LoRA? This depends on your LoRA's complexity. For example, when I training the spritesheet LoRA, even though I used 100 images, I didn't include buildings in the training data, so this LoRA doesn't work for buildings. So think about your LoRA's use cases and add examples for as many use cases as possible to improve quality. What are paired images and how to train LoRAs for image-editing? When training LoRAs for image editing on fal, we call each edit example paired images - one with _start suffix, one with _end suffix. For example, if you're training a background remove LoRA, the unedited original photo will be your '_start' image. The image with background removed will be the '_end' image. Simply put: images we want to edit or use as reference get _start, target images we want to achieve get '_end'. Important: save both images with the same name. Like image332_start.jpg and image332_end.jpg. This way the system knows which images pair together. What about training LoRAs for models with multiple image inputs? Same logic. We still use _start and _end suffixes, but with one difference. Since there are multiple input images, we can number them: _start, _start1, _start2. Example: start images, 1st image = Woman portrait (image35_start.jpg) 2nd image = Glasses photo (image35_start1.jpg) 3rd image = Hat photo (image35_start2.jpg) Output image = portrait of woman wearing glasses and hat (image35_end.jpg) Can we do more detailed captioning? Yes. Similarly, you can improve training quality by creating a txt file for each set with the caption inside. Example: create image35.txt and write: 'Recreate the image by putting the glasses from the second image and the hat from the third image on the woman in the first image.' What are Steps? How many should I use? What's Learning Rate? Steps determines how many times the model sees and processes your training data (your images). Each step, the model learns a bit more. But as steps increase, so does the risk of overfitting. So there's no real default. But for a simpler LoRA with 20 paired images, 1000 steps is ideal. Here's a metaphor for the Steps and Learning Rate relationship: Imagine you have a balloon. Our goal is to inflate it to the optimal size. Steps = How many times we blow into the balloon Learning rate = How hard we blow each time If we blow too softly, we need to blow many more times. If we blow too hard, we risk popping it quickly and can't reach optimal size. Of course training won't explode, but it won't work as intended because it wasn't trained optimally. Training's done, now what? Once training's complete, you'll have a safetensors file. Every model you train on fal has a LoRA inference endpoint. In that inference, add your safetensors file link to the LoRA url input, and you can use your LoRA. Thanks for the read! The workflow in the video: If I forgot anything, let me know in the replies.

As I promised yesterday, I'll briefly explain LoRA training and share a workflow I made so you can do it quickly. First, let me answer a very common question: 'Why train LoRAs when we have such advanced models?' Even though we have incredibly advanced models now (like NBP), we still can't always get them to do specific things we want. Simplest example: the spritesheet LoRA I made the other day. I generated 1000 images with Nano Banana and only 100 were what I wanted. The LoRA I trained using those 100 images gives me nearly 100% consistent results. Second point is cost and speed. With LoRA, we can cut costs by 4-5x. And while doing that, we're generating 4-5x faster. How many images do you need for a good LoRA? This depends on your LoRA's complexity. For example, when I training the spritesheet LoRA, even though I used 100 images, I didn't include buildings in the training data, so this LoRA doesn't work for buildings. So think about your LoRA's use cases and add examples for as many use cases as possible to improve quality. What are paired images and how to train LoRAs for image-editing? When training LoRAs for image editing on fal, we call each edit example paired images - one with _start suffix, one with _end suffix. For example, if you're training a background remove LoRA, the unedited original photo will be your '_start' image. The image with background removed will be the '_end' image. Simply put: images we want to edit or use as reference get _start, target images we want to achieve get '_end'. Important: save both images with the same name. Like image332_start.jpg and image332_end.jpg. This way the system knows which images pair together. What about training LoRAs for models with multiple image inputs? Same logic. We still use _start and _end suffixes, but with one difference. Since there are multiple input images, we can number them: _start, _start1, _start2. Example: start images, 1st image = Woman portrait (image35_start.jpg) 2nd image = Glasses photo (image35_start1.jpg) 3rd image = Hat photo (image35_start2.jpg) Output image = portrait of woman wearing glasses and hat (image35_end.jpg) Can we do more detailed captioning? Yes. Similarly, you can improve training quality by creating a txt file for each set with the caption inside. Example: create image35.txt and write: 'Recreate the image by putting the glasses from the second image and the hat from the third image on the woman in the first image.' What are Steps? How many should I use? What's Learning Rate? Steps determines how many times the model sees and processes your training data (your images). Each step, the model learns a bit more. But as steps increase, so does the risk of overfitting. So there's no real default. But for a simpler LoRA with 20 paired images, 1000 steps is ideal. Here's a metaphor for the Steps and Learning Rate relationship: Imagine you have a balloon. Our goal is to inflate it to the optimal size. Steps = How many times we blow into the balloon Learning rate = How hard we blow each time If we blow too softly, we need to blow many more times. If we blow too hard, we risk popping it quickly and can't reach optimal size. Of course training won't explode, but it won't work as intended because it wasn't trained optimally. Training's done, now what? Once training's complete, you'll have a safetensors file. Every model you train on fal has a LoRA inference endpoint. In that inference, add your safetensors file link to the LoRA url input, and you can use your LoRA. Thanks for the read! The workflow in the video: If I forgot anything, let me know in the replies.

ilker

15,192 Aufrufe • vor 7 Monaten

Hi. The last 2 weeks I’ve been building an iOS watch only wallet for signing $KAS transactions offline using an M5Stack CoreS3 running KasSigner software. Below I demonstrate building a kaspa tx locally, signing it offline, and then broadcasting it to the network. I am still building it and have not had the code audited yet so this should be considered only a proof of concept. The $KAS community has very few options for good wallets so I decided to build one that has everything I could ever want with a simple, sharp UI, privacy, and UTXO control as the most basic foundation, and not a “future roadmap update” pushed out by a company that has 100,000 other tokens they have to dedicate time and resources to. I also plan to build covenants support at some point, but for now I want to have a really good app for signing kaspa txs offline. Also if anyone wants to offer me tips, support, or guidance my DMs are open. I’m also doing this for free in my spare time so if anyone is interested in donating you can at brt25.kas or kaspa:qqpzpn5e7enn2ylfdxvlwtm3829gn6j9z9dnnmcsw5arkgnurktty6ulgzkfk Hope you guys like it and I will be releasing it on GitHub as soon as I clean up the code.

Hi. The last 2 weeks I’ve been building an iOS watch only wallet for signing $KAS transactions offline using an M5Stack CoreS3 running KasSigner software. Below I demonstrate building a kaspa tx locally, signing it offline, and then broadcasting it to the network. I am still building it and have not had the code audited yet so this should be considered only a proof of concept. The $KAS community has very few options for good wallets so I decided to build one that has everything I could ever want with a simple, sharp UI, privacy, and UTXO control as the most basic foundation, and not a “future roadmap update” pushed out by a company that has 100,000 other tokens they have to dedicate time and resources to. I also plan to build covenants support at some point, but for now I want to have a really good app for signing kaspa txs offline. Also if anyone wants to offer me tips, support, or guidance my DMs are open. I’m also doing this for free in my spare time so if anyone is interested in donating you can at brt25.kas or kaspa:qqpzpn5e7enn2ylfdxvlwtm3829gn6j9z9dnnmcsw5arkgnurktty6ulgzkfk Hope you guys like it and I will be releasing it on GitHub as soon as I clean up the code.

₿RT 𐤊 🐈📈

23,811 Aufrufe • vor 1 Monat

Holy Sh*t... You can now share your localhost with anyone using only VS Code! 🤯 VS Code has a built-in Ports feature that lets you expose your local server to the internet, no need for third-party tools. Here’s how to use it: 1) Open the Ports tab in the terminal panel 2) Click Forward a Port 3) Enter your local port number (mine was 5173, yours could be 3000, 3001, etc.) 4) VS Code will generate a link, right-click and select Make Public 5) Copy the link and share it with anyone Perfect for demos, testing on other devices, or sharing work with clients and teammates :) Watch the video for a step-by-step demo 🎥 Follow me for more amazing AI, Coding & Web Dev insights 💎

Holy Sh*t... You can now share your localhost with anyone using only VS Code! 🤯 VS Code has a built-in Ports feature that lets you expose your local server to the internet, no need for third-party tools. Here’s how to use it: 1) Open the Ports tab in the terminal panel 2) Click Forward a Port 3) Enter your local port number (mine was 5173, yours could be 3000, 3001, etc.) 4) VS Code will generate a link, right-click and select Make Public 5) Copy the link and share it with anyone Perfect for demos, testing on other devices, or sharing work with clients and teammates :) Watch the video for a step-by-step demo 🎥 Follow me for more amazing AI, Coding & Web Dev insights 💎

Ram Maheshwari

43,130 Aufrufe • vor 1 Monat

📘 Holiday read! From Software Engineer to AI Environment Architect 🚀 Tldr of our blog: We see an exciting future where engineers 👩‍💻 won’t stop coding — but the highest leverage shifts to designing the environments 🛝 where AI can think, build, and evolve. 🎬 Demo: Inspired by opinions from Andrej Karpathy Richard Sutton, our newly built framework Vortex shows this in a concrete action: by architecting the right environment in LLM serving systems, an agent from OpenHands can generate and implement new Sparse Attention algorithms on SGLang in a single run and deliver up to 4× ⏩ gains — work that normally takes an ML-systems engineer weeks. - In the short term, these environments let AI agents contribute meaningfully to real engineering work today. - In the long term, they become the playgrounds where future agents learn to surpass today’s limitations 💖. Get too excited by the demo, write a blogpost before the holidays with my great students chen zhuoming Yang Zhou who built it: Vortex code: Vortex doc: #MLOps #AIAgents #SystemsEngineering #AIInfrastructure #OpenSourceAI #AIOps #AIFrameworks #SparseAttention #AIResearch

📘 Holiday read! From Software Engineer to AI Environment Architect 🚀 Tldr of our blog: We see an exciting future where engineers 👩‍💻 won’t stop coding — but the highest leverage shifts to designing the environments 🛝 where AI can think, build, and evolve. 🎬 Demo: Inspired by opinions from Andrej Karpathy Richard Sutton, our newly built framework Vortex shows this in a concrete action: by architecting the right environment in LLM serving systems, an agent from OpenHands can generate and implement new Sparse Attention algorithms on SGLang in a single run and deliver up to 4× ⏩ gains — work that normally takes an ML-systems engineer weeks. - In the short term, these environments let AI agents contribute meaningfully to real engineering work today. - In the long term, they become the playgrounds where future agents learn to surpass today’s limitations 💖. Get too excited by the demo, write a blogpost before the holidays with my great students chen zhuoming Yang Zhou who built it: Vortex code: Vortex doc: #MLOps #AIAgents #SystemsEngineering #AIInfrastructure #OpenSourceAI #AIOps #AIFrameworks #SparseAttention #AIResearch

Beidi Chen

41,039 Aufrufe • vor 9 Monaten