RTX A6000, lists 48Gb video memory...
Soooo jealous....
dont be its ampere .. i have 2 of them .. i much rather have 2 6000 pro nowadays .. much much faster
MrDragonFox
MrDragonFox
AI & ML interests
llm + audio i/o, (un)alignment
Recent Activity
liked
a model
17 days ago
lilylilith/AnyPose
upvoted
a
collection
29 days ago
sam-audio
upvoted
an
article
29 days ago
AI Labs Must Resist Age Verification
Organizations
reacted to
prithivMLmods's
post with 🚀
9 months ago
Post
2028
Dropping downstream tasks using newly initialized parameters and weights supports domain-specific image classification post-training, based on the SigLIP-2 models: Patch-16/224, Patch-16/256, and Patch-32/256. For more details, please refer to the respective model cards : 🤗
+ watermark detection : prithivMLmods/Watermark-Detection-SigLIP2
+ resisc45 : prithivMLmods/RESISC45-SigLIP2
+ pacs dg : prithivMLmods/PACS-DG-SigLIP2
+ 3d printed or not : prithivMLmods/3D-Printed-Or-Not-SigLIP2
+ formula or text : prithivMLmods/Formula-Text-Detection
- explicit content patch16 256 : prithivMLmods/siglip2-x256-explicit-content
- explicit content patch32 256 : prithivMLmods/siglip2-x256p32-explicit-content
> SigLIP2 Content Filters 042025 Final : https://huggingface.co/collections/prithivMLmods/siglip2-content-filters-04202-final-680fe4aa1a9d589bf2c915ff
> SigLIP2 : google/siglip2-67b5dcef38c175486e240107
> SigLIP2 Multilingual Vision-Language Encoders : https://arxiv.org/pdf/2502.14786
+ watermark detection : prithivMLmods/Watermark-Detection-SigLIP2
+ resisc45 : prithivMLmods/RESISC45-SigLIP2
+ pacs dg : prithivMLmods/PACS-DG-SigLIP2
+ 3d printed or not : prithivMLmods/3D-Printed-Or-Not-SigLIP2
+ formula or text : prithivMLmods/Formula-Text-Detection
Categorizing Un-Safe Content :- explicit content patch16 256 : prithivMLmods/siglip2-x256-explicit-content
- explicit content patch32 256 : prithivMLmods/siglip2-x256p32-explicit-content
Collection :> SigLIP2 Content Filters 042025 Final : https://huggingface.co/collections/prithivMLmods/siglip2-content-filters-04202-final-680fe4aa1a9d589bf2c915ff
> SigLIP2 : google/siglip2-67b5dcef38c175486e240107
> SigLIP2 Multilingual Vision-Language Encoders : https://arxiv.org/pdf/2502.14786
posted
an
update
9 months ago
Post
4550
as a few of you know - i am working on a rather more elaborate-tts that can produce more interesting sounds in context of rp
early sneak peak is here -
MrDragonFox/mOrpheus_3B-1Base_early_preview-v1-25000
its based on orpheus - but really the model is irrelevant as i focus mostly on data augmentation / prep / pipelineing - its just the way to show progress
should be able to express fine even in a sfw context
probably the last release for a few weeks as i go back to the data pipeline and improve there ..
in the mean time, please do test and report problems or enjoyable generations you found - we have a growing discord community and i love to see what you get out of that early release !
(small colab is provided on the model page if you dont have the gpu to run that your self)
early sneak peak is here -
MrDragonFox/mOrpheus_3B-1Base_early_preview-v1-25000
its based on orpheus - but really the model is irrelevant as i focus mostly on data augmentation / prep / pipelineing - its just the way to show progress
should be able to express fine even in a sfw context
probably the last release for a few weeks as i go back to the data pipeline and improve there ..
in the mean time, please do test and report problems or enjoyable generations you found - we have a growing discord community and i love to see what you get out of that early release !
(small colab is provided on the model page if you dont have the gpu to run that your self)
posted
an
update
9 months ago
Post
5622
yet a other audio datasets pre classified for events + audio aestetics
this time for german - 680h sampled from emilia yodas
timestamps for asr training or other fancier things available as nc in the raw repo
MrDragonFox/DE_Emilia_Yodas_680h
cc by 4.0 as by emilia yodas
raw events / transcriptions are cc by NC 4.0
MrDragonFox/DE_Emilia_Yodas_680h_raw_timestamps
the coming days i should push about 600h english + some japanese too same format
this time for german - 680h sampled from emilia yodas
timestamps for asr training or other fancier things available as nc in the raw repo
MrDragonFox/DE_Emilia_Yodas_680h
cc by 4.0 as by emilia yodas
raw events / transcriptions are cc by NC 4.0
MrDragonFox/DE_Emilia_Yodas_680h_raw_timestamps
the coming days i should push about 600h english + some japanese too same format
posted
an
update
10 months ago
Post
2152
did a small emotive classified test dataset for all the tts tuners out there
MrDragonFox/Elise
3h total mit - single speaker voice
dataset is a copy of an existing one just added the emotional tags over 1200 samples - should be good enough to test if emotional tags stick in your finetune
MrDragonFox/Elise
3h total mit - single speaker voice
dataset is a copy of an existing one just added the emotional tags over 1200 samples - should be good enough to test if emotional tags stick in your finetune
- 1 reply
replied to
Reality123b's
post
11 months ago
bro, im not that popular. its ok to do this ig?
also, no, im not saying openai's products are bad. nor im trying to like offend ANYONE or ANY company or organization. im just trying to promote my product
my problem isnt with you trying to promote a product - you asking if you should opensource something or not .. without showing anything working and state its not clickbait ..
show the goods man ..
otherwise noone cares - not how marketing works .. too many snakeoil sales men in that industry ... if you have something show it off .. if not .. well you know where this goes
replied to
Reality123b's
post
11 months ago
the point you are mistaken here is that i dont care if he opensources it or not .. its the engagement farming for no reason while implying its not clickbait - but you do you
replied to
Reality123b's
post
11 months ago
this is not clickbait .. if you want to opensource it .. you opensource it .. if not you dont .. its that simple - there are other approaches out there to this already
if that's not for farming what did you post it for ?
is that with ddr4 or 5 ?
with 250g ram used ^^ probably running it at a 2 bit quant .
reacted to
danielhanchen's
post with 🔥
about 1 year ago
Post
1560
Vision finetuning is in 🦥Unsloth! You can now finetune Llama 3.2, Qwen2 VL, Pixtral and all Llava variants up to 2x faster and with up to 70% less VRAM usage! Colab to finetune Llama 3.2: https://colab.research.google.com/drive/1j0N4XTY1zXXy7mPAhOC1_gMYZ2F2EBlk?usp=sharing
- 1 reply
replied to
davanstrien's
post
over 1 year ago
96% to wikipedia - i love the idea - but the similarity estimation is far off
original dataset
https://huggingface.co/datasets/kalomaze/Opus_Instruct_25k
replied to
DrHouseFan-315's
post
over 1 year ago
sure and who is gonna pay for that ?
reacted to
merve's
post with 🤗
over 1 year ago
Post
6138
Fine-tune Florence-2 on any task 🔥
Today we release a notebook and a walkthrough blog on fine-tuning Florence-2 on DocVQA dataset @andito @SkalskiP
Blog: https://huggingface.co/blog 📕
Notebook: https://colab.research.google.com/drive/1hKDrJ5AH_o7I95PtZ9__VlCTNAo1Gjpf?usp=sharing 📖
Florence-2 is a great vision-language model thanks to it's massive dataset and small size!
This model requires conditioning through task prefixes and it's not as generalist, requiring fine-tuning on a new task, such as DocVQA 📝
We have fine-tuned the model on A100 (and one can also use a smaller GPU with smaller batch size) and saw that model picks up new tasks 🥹
See below how it looks like before and after FT 🤩
Play with the demo here andito/Florence-2-DocVQA 🏄♀️
Today we release a notebook and a walkthrough blog on fine-tuning Florence-2 on DocVQA dataset @andito @SkalskiP
Blog: https://huggingface.co/blog 📕
Notebook: https://colab.research.google.com/drive/1hKDrJ5AH_o7I95PtZ9__VlCTNAo1Gjpf?usp=sharing 📖
Florence-2 is a great vision-language model thanks to it's massive dataset and small size!
This model requires conditioning through task prefixes and it's not as generalist, requiring fine-tuning on a new task, such as DocVQA 📝
We have fine-tuned the model on A100 (and one can also use a smaller GPU with smaller batch size) and saw that model picks up new tasks 🥹
See below how it looks like before and after FT 🤩
Play with the demo here andito/Florence-2-DocVQA 🏄♀️