LoRA Training Tutorial｜TensorArt Feature Update✨

[@apa_channel_]

there are many models in tensor-art but there are very few base models for making lora. can you add a custom base model menu so that users can be creative, can choose from more and use their own settings.

[@TensorArt]

hello, sure~ will optimize it soon!

[@apa_channel_]

@@TensorArt when is it implemented, I've been waiting for a long time sir

[@TensorArt]

@@apa_channel_ Hey there! We're excited to let you know that we've got some fresh additions, including the pony model. Can't wait for you to check it out!

[@apa_channel_]

@@TensorArt for the base model pony xl option, it is wrongly detecting the image size, so 512 instead of 1024. can you fix it?

[@apa_channel_]

@@TensorArt Yes I've seen the new pony options, I'm also waiting for hassanblend v1.4 and bra v5.

[@WowomboCocombo]

awesome explanation. thanks!

[@nowxrlds446]

i love this website!! thank you

[@DeAdLyRuShN]

Would like to learn more on the upload process after training. With advice on best settings etc.

[@TensorArt]

It is on the way, keep following!

[@SimplyTiffany2009]

me to idk how to use it after i trained it

[@mrtim6479]

When I tried this my images all came out blank. No idea how to get support

[@MikeBNumba6]

After the training is complete, its asking me to host a model. Unsure how that process works

[@TensorArt]

If you want to use the lora to create images, you have to upload it on our website through your profile~

[@MikeBNumba6]

@@TensorArt thanks again for the quick reply

[@ImmortalShiro]

Asking to host a model is basically just making a model project in TensorArt. Just create one, then go back to the training page and select your trained model then published it to the project you've recently created.

[@TensorArt]

Our new video is about model host, you can watch it.

[@D87ify]

Question If I have an 500 image’s of anime character what steps and Epoch I need?

[@experimentalpunksurfer]

TensorArt, I wonder if it would be possible to upgrade to the pro plan multiple times for 24 hours only. I am usually busy with work and can only do production activities on one day off.

[@TensorArt]

Do you mean you already paid for a one day pro but want to buy it more?

[@experimentalpunksurfer]

@@TensorArt No, I have not purchased a one-day professional plan yet. Because it is a one-time opportunity and I would like to purchase it when I can afford a longer period of time.

[@TensorArt]

@@experimentalpunksurfer Okay, we will take this into consideration

[@experimentalpunksurfer]

@@TensorArt Thank you for hearing me out.

[@leandrostasi]

do I have to download and then upload it back? do I need to do it with all the epochs, or just the last one? I got a big file when downloading an SDXL, like 10 times what a SD1.5 lora file used to weight, is that all right?

[@TensorArt]

You need to download it and upload it now, we will skip the download step in the future～One is enough but more iz better.The file weight is based on your images number and parameters, usually below 1GB.

[@Brainomaniak]

It says "Sorry, after system review, the image you uploaded may contain content that is not suitable for model training. Thank you for understanding." But it's just an anime character?

[@Koonezone]

Hello there First of all thank you for you ai And I have question I want make character anime but I don't want the clothes change I want make many style and faceses with one clothes but always I make new style or new action the clothes change what I must to do ?

[@TensorArt]

Thank you for reaching out with your question. A professional solution would be to explore the AI Tools section for some clothing change tools. If someone has shared their workflow, you can download their clothing change workflow. You simply need to upload the photo of the clothing you desire and modify the prompts to maintain the same clothing while changing the character’s actions. Alternatively, you could open the workspace, generate an image of the character with the desired clothing, and then copy the seed. Save this image, and then upload it to controlnet’s openpose or inpaint. By using the same seed and modifying the prompts for actions, you should be able to achieve the desired result. I hope this is helpful to you. If you encounter any issues, please feel free to share your experience in our Discord community for further assistance.

[@boomorningstar3061]

It asks me to host model, upload files? Don't have pictures of the thing I'm creating because it not published yet? Then it asks me choose a base model?😮

[@KiritoxNemesis]

Can you make a picture of a real person? I tried it and why did the result come out like an anime?

[@teamsleepnine]

what is trigger words? also when can we subscribed using apple via telco on the apps,thanks!

[@TensorArt]

Trigger words are those mostly can describe the feature of your lora. Add it in your prompts will trigger to use the lora. btw we always support Apple subscription payments within the app~

[@teamsleepnine]

@@TensorArt whenever I try to subscribed it said "no corresponding product found"

[@TensorArt]

@@teamsleepnine do you mean you subscribed TensorArt in App store?

[@teamsleepnine]

yes, from from tensor phone apps @@TensorArt

[@TensorArt]

ok, we will check it.

[@johnmullins5258]

What causes this message? The format of the model file is wrong, needs to be re-uploaded. I uploaded my images then selected realistic for model theme then I chose Juggernaut XL as the base model but when the training was completed I got that message, and I did what this tutorial said to do.

[@TensorArt]

You can join our Discord community and post this question in the group messages. We have staff members who will assist you.

[@johnmullins5258]

@@TensorArt I hate Discord and don't use it. Your video tutorials should cover things like this and how to fix them.

[@TensorArt]

@@johnmullins5258 Please send me your personal ID. You can copy the URL link from your profile page and provide me with the numerical part.

[@johnmullins5258]

@@TensorArt I sent you my url on KZbin in a comment but the idiots at KZbin apparently deleted that comment because of the url link, so I will send you my url on your X aka twitter account.

[@TensorArt]

@@johnmullins5258 ok

[@morganfreeman417]

why after it finishing online training, no epoch pictures show ? and it just blank, all epoch not working ... but it count my credits ....

[@TensorArt]

Thank you for your patience and for bringing this issue to our attention. It's possible that the problem you're experiencing could be due to a network issue. I would suggest trying to refresh the webpage a few more times. If this doesn't work, you may consider switching to a different web browser to isolate the problem. If you continue to encounter difficulties after trying these steps, please let us know. We're here to help in any way we can. Once again, thank you for your understanding and patience.

[@datehive7604]

Please show full tutorial, pls do not make a lazy video. It keep getting gernerate failed with my lora model file lora-000009 (1).safetensors is deploying, "it take 10-15 minute to deploy. It's been two days... There seems no any documentation about this

[@TensorArt]

You can join our Discord community to report the problem you met. More videos are on the way.

[@tuhinbiswas98]

my train lora will be private???? or eveyone can use it???? i am asking for free acount?

[@TensorArt]

You can set your lora will be opened for all or not when you are uploading your lora.It depends on you~

[@luismiguel7256]

How do I update my lora?

[@TensorArt]

You can update your model by clicking "+ New Version" on the main page of your model.

[@heitorb2460]

Hi, I ran a couple of training sessions following your instructions (spend a lot of credits doing so). Published one of the Epochs (as private). How do I use now? Do i have to make it public so I myself can use the one I created? I was expecting that on the creation page I could click Add LoRa and select it somehow.

[@TensorArt]

You certainly can use your private lora! You can click into your private lora page and clik run and you can use it～

[@testa-x7z]

how do you publish the epochs as private?

[@ziogregory4799]

You cant mix on tensor unless you pay!

[@ziogregory4799]

Has TensorArt changed their policies on free use? Can't load mix images without warning sign to upgrade to vip pro?

[@TensorArt]

@@ziogregory4799 this isn't included in pro priority.

[@ucanhta1862]

How to host a checkpoint model? I can only train lora model

[@TensorArt]

you can check the model hosting tutorial in another video we uploaded. the method is the same.

[@davedmuttart6789]

This is a particularly poor training video. It skips certain steps, it quickly glosses over other essential steps and fails to properly explain. The screen we are being instructed on how to use, is minimized in the center of the screen. All in all, it's part of the reason why I have spent so much time and frustration trying to get TA to work instead of generating images.

[@TensorArt]

Thank you for your criticism and suggestions. We've updated our content with the latest professional training tutorial. You can find it here: kzbin.info/www/bejne/oHPCdad_Ypidhq8. We hope you find it helpful!