Train a Llama model from scratch
•
45
Noa Roggendorff
nroggendorff
AI & ML interests
None yet
Recent Activity
updated
a Space
7 minutes ago
nroggendorff/nroggendorff
updated
a model
32 minutes ago
nroggendorff/smallama
updated
a Space
about 1 hour ago
nroggendorff/train-llama
Articles
Organizations
nroggendorff's activity
posted
an
update
about 11 hours ago
Post
206
I've done it.
~~well, some of it~~
~~well, some of it~~
- 2 replies
replied to
their
post
5 days ago
merges.txt :spinnyhat:
Reacted to
luigi12345's
post with 👀
13 days ago
Post
2244
Best Debug Prompt
You are a frustrated user who has tested this application extensively. Your job is to list EVERY possible way this app could completely break or become unusable.
For each potential failure:
1. What would make you say "This app is totally broken!"?
2. What exact steps did you take when it broke?
3. What did you see on your screen when it broke?
4. How angry would this make a typical user (1-10)?
5. What would you expect the app to do instead?
Think about:
- What happens if you click buttons really fast?
- What if your internet is slow/disconnected?
- What if you upload weird files/images?
- What if you try to break the app on purpose?
- What if multiple people use it at once?
- What if you use it on mobile/tablet?
- What if you refresh/navigate while it's working?
- What if you paste invalid inputs?
- What if you upload HUGE files?
- What if you leave it running overnight?
Don't worry about being technical - just describe what you saw break as a user.
Format each issue like:
ISSUE #1: [Brief angry user description]
- STEPS TO BREAK IT: [Exactly what you did]
- WHAT HAPPENED: [What you saw]
- ANGER LEVEL: [1-10]
- EXPECTED: [What should happen]
Keep going until you've found every possible way to break this app from a user's perspective!
After outpuiting the list, accoring to the list optmiced Composer edit block to fix the ones severe that make sense to adjust accoirng to gradio limitations and current usage target )dont suppose we need unecessary funcitons)
You are a frustrated user who has tested this application extensively. Your job is to list EVERY possible way this app could completely break or become unusable.
For each potential failure:
1. What would make you say "This app is totally broken!"?
2. What exact steps did you take when it broke?
3. What did you see on your screen when it broke?
4. How angry would this make a typical user (1-10)?
5. What would you expect the app to do instead?
Think about:
- What happens if you click buttons really fast?
- What if your internet is slow/disconnected?
- What if you upload weird files/images?
- What if you try to break the app on purpose?
- What if multiple people use it at once?
- What if you use it on mobile/tablet?
- What if you refresh/navigate while it's working?
- What if you paste invalid inputs?
- What if you upload HUGE files?
- What if you leave it running overnight?
Don't worry about being technical - just describe what you saw break as a user.
Format each issue like:
ISSUE #1: [Brief angry user description]
- STEPS TO BREAK IT: [Exactly what you did]
- WHAT HAPPENED: [What you saw]
- ANGER LEVEL: [1-10]
- EXPECTED: [What should happen]
Keep going until you've found every possible way to break this app from a user's perspective!
After outpuiting the list, accoring to the list optmiced Composer edit block to fix the ones severe that make sense to adjust accoirng to gradio limitations and current usage target )dont suppose we need unecessary funcitons)
Reacted to
MonsterMMORPG's
post with 🔥
19 days ago
Post
4532
Hunyuan3D-1 - SOTA Open Source Text-to-3D and Image-to-3D - 1-Click Install and use both Locally on Windows and on Cloud - RunPod and Massed Compute
Automatic Installers
Works amazing on 24 GB GPUs
Files > https://www.patreon.com/posts/115412205
So what is Hunyuan3D-1
Official repo : https://github.com/tencent/Hunyuan3D-1
On Hugging Face : tencent/Hunyuan3D-1
Tencent Hunyuan3D-1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation
Abstract
While 3D generative models have greatly improved artists' workflows, the existing diffusion models for 3D generation suffer from slow generation and poor generalization. To address this issue, we propose a two-stage approach named Hunyuan3D-1.0 including a lite version and a standard version, that both support text- and image-conditioned generation.
In the first stage, we employ a multi-view diffusion model that efficiently generates multi-view RGB in approximately 4 seconds. These multi-view images capture rich details of the 3D asset from different viewpoints, relaxing the tasks from single-view to multi-view reconstruction. In the second stage, we introduce a feed-forward reconstruction model that rapidly and faithfully reconstructs the 3D asset given the generated multi-view images in approximately 7 seconds. The reconstruction network learns to handle noises and in-consistency introduced by the multi-view diffusion and leverages the available information from the condition image to efficiently recover the 3D structure.
Our framework involves the text-to-image model, i.e., Hunyuan-DiT, making it a unified framework to support both text- and image-conditioned 3D generation. Our standard version has 3x more parameters than our lite and other existing model. Our Hunyuan3D-1.0 achieves an impressive balance between speed and quality, significantly reducing generation time while maintaining the quality and diversity of the produced assets.
Automatic Installers
Works amazing on 24 GB GPUs
Files > https://www.patreon.com/posts/115412205
So what is Hunyuan3D-1
Official repo : https://github.com/tencent/Hunyuan3D-1
On Hugging Face : tencent/Hunyuan3D-1
Tencent Hunyuan3D-1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation
Abstract
While 3D generative models have greatly improved artists' workflows, the existing diffusion models for 3D generation suffer from slow generation and poor generalization. To address this issue, we propose a two-stage approach named Hunyuan3D-1.0 including a lite version and a standard version, that both support text- and image-conditioned generation.
In the first stage, we employ a multi-view diffusion model that efficiently generates multi-view RGB in approximately 4 seconds. These multi-view images capture rich details of the 3D asset from different viewpoints, relaxing the tasks from single-view to multi-view reconstruction. In the second stage, we introduce a feed-forward reconstruction model that rapidly and faithfully reconstructs the 3D asset given the generated multi-view images in approximately 7 seconds. The reconstruction network learns to handle noises and in-consistency introduced by the multi-view diffusion and leverages the available information from the condition image to efficiently recover the 3D structure.
Our framework involves the text-to-image model, i.e., Hunyuan-DiT, making it a unified framework to support both text- and image-conditioned 3D generation. Our standard version has 3x more parameters than our lite and other existing model. Our Hunyuan3D-1.0 achieves an impressive balance between speed and quality, significantly reducing generation time while maintaining the quality and diversity of the produced assets.
posted
an
update
20 days ago
Post
2226
I still think whitespace in tokenizers are so dumb.
Congrats, you just doubled your vocab size for no reason.
Congrats, you just doubled your vocab size for no reason.
-
- 3 replies
replied to
their
post
22 days ago
The model I made isn't very good by the way, it's only set for one epoch.
replied to
their
post
23 days ago
replied to
their
post
24 days ago
They will charge you a 10$ verification fee (so don't lock until that happens), but that cancels after a few days.
posted
an
update
24 days ago
Post
1820
Did you guys know that if you try to link a prepaid card to huggingface it won't work, but then if you press the button again it links anyway? Then you can lock the card (deny any charges), and get resources for free? You're welcome :P
-
-
- 4 replies
replied to
MichaelBoll's
post
about 1 month ago
no
replied to
their
post
about 1 month ago
posted
an
update
about 1 month ago
Post
2105
she assert on my device until i give up AHAHEGHFDGHJHASUFSHD
-
-
- 3 replies
Reacted to
their
post with 🚀
about 2 months ago
Post
2645
When huggingface patches this, I'm going to be really sad, but in the meantime, here you go:
When AutoTrain creates a new space to train your model, it does so via the huggingface API. If you modify the code so that it includes a premade README.md file, you can add these two lines:
This will tell huggingface to listen for the iframe on your port, instead of the one autotrain is actually hosting on, and because startup time isn't charged, you get the product for free. (you can take this even further by switching compute type to A100 or something)
When AutoTrain creates a new space to train your model, it does so via the huggingface API. If you modify the code so that it includes a premade README.md file, you can add these two lines:
---
app_port: 8080 # or any integer besides 7860 that's greater than 2 ** 10
startup_duration_timeout: 350m
---This will tell huggingface to listen for the iframe on your port, instead of the one autotrain is actually hosting on, and because startup time isn't charged, you get the product for free. (you can take this even further by switching compute type to A100 or something)
-
- 1 reply
posted
an
update
about 2 months ago
Post
2645
When huggingface patches this, I'm going to be really sad, but in the meantime, here you go:
When AutoTrain creates a new space to train your model, it does so via the huggingface API. If you modify the code so that it includes a premade README.md file, you can add these two lines:
This will tell huggingface to listen for the iframe on your port, instead of the one autotrain is actually hosting on, and because startup time isn't charged, you get the product for free. (you can take this even further by switching compute type to A100 or something)
When AutoTrain creates a new space to train your model, it does so via the huggingface API. If you modify the code so that it includes a premade README.md file, you can add these two lines:
---
app_port: 8080 # or any integer besides 7860 that's greater than 2 ** 10
startup_duration_timeout: 350m
---This will tell huggingface to listen for the iframe on your port, instead of the one autotrain is actually hosting on, and because startup time isn't charged, you get the product for free. (you can take this even further by switching compute type to A100 or something)
-
- 1 reply
posted
an
update
about 2 months ago
Post
1518
pretty much all of the values in the llama training post are placeholders, so if you dont get a desireable result tweak and tweak and tweak. it took months to get smallama to do anything
Reacted to
zamal's
post with 🤗
about 2 months ago
Post
1938
🚀 New Model Release: zamal/Molmo-7B-GPTQ-4bit 🚀
Hello lovely community,
zamal/Molmo-7B-GPTQ-4bit model is now available for all! This model has been highly quantized, reducing its size by almost six times. It now occupies significantly less space and vRAM, making it perfect for deployment on resource-constrained devices without compromising performance.
Now we get:
Efficient Performance: Maintains high accuracy while being highly quantized.
Reduced Size: The model size is reduced by nearly six times, optimizing storage and memory usage.
Versatile Application: Ideal for integrating a powerful visual language model into various projects particularly multi rag chains.
Check it out!
Hello lovely community,
zamal/Molmo-7B-GPTQ-4bit model is now available for all! This model has been highly quantized, reducing its size by almost six times. It now occupies significantly less space and vRAM, making it perfect for deployment on resource-constrained devices without compromising performance.
Now we get:
Efficient Performance: Maintains high accuracy while being highly quantized.
Reduced Size: The model size is reduced by nearly six times, optimizing storage and memory usage.
Versatile Application: Ideal for integrating a powerful visual language model into various projects particularly multi rag chains.
Check it out!
- 1 reply