already Supra3?
Not yet, we're just announcing a plan, that would take ~1-2 months
๐๏ธ Building on HF
AxionLab
AxionLab-official
AI & ML interests
Owner of SupraLabs and iGPU Lover
Recent Activity
new activity about 3 hours ago
wop/littlechat-5m-instruct:Samples repliedto LH-Tech-AI's post about 3 hours ago
Hey community and sponsors!
We are announcing the Supra3 family with four core SLM models:
- Supra3 Flash Lite: 25M parameters, ~60B pretraining tokens
- Supra3 Flash: 50M parameters, ~100B pretraining tokens
- Supra3 Pro: 75M parameters, ~150B pretraining tokens
- Supra3 Ultra: 100M parameters, ~200B pretraining tokens
For Supra3 Pro and Ultra, we search for sponsors who give us free access to compute like RTX 5090 32GB or so.
We estimate the total cost of the pro and ultra models at around $600.
For Supra3 Flash Lite and Flash, we do not need sponsors.
If anyone would apply for helping us, we would be really thankful and this person would get early access to new modele, insider information, credit and more!
Contact: here or on discord: lh_tech_ai reacted to LH-Tech-AI's post with ๐ about 3 hours ago
Hey community and sponsors!
We are announcing the Supra3 family with four core SLM models:
- Supra3 Flash Lite: 25M parameters, ~60B pretraining tokens
- Supra3 Flash: 50M parameters, ~100B pretraining tokens
- Supra3 Pro: 75M parameters, ~150B pretraining tokens
- Supra3 Ultra: 100M parameters, ~200B pretraining tokens
For Supra3 Pro and Ultra, we search for sponsors who give us free access to compute like RTX 5090 32GB or so.
We estimate the total cost of the pro and ultra models at around $600.
For Supra3 Flash Lite and Flash, we do not need sponsors.
If anyone would apply for helping us, we would be really thankful and this person would get early access to new modele, insider information, credit and more!
Contact: here or on discord: lh_tech_aiOrganizations
replied to LH-Tech-AI's post about 3 hours ago
reacted to LH-Tech-AI's post with ๐๐ฅ about 3 hours ago
Post
192
Hey community and sponsors!
We are announcing the Supra3 family with four core SLM models:
- Supra3 Flash Lite: 25M parameters, ~60B pretraining tokens
- Supra3 Flash: 50M parameters, ~100B pretraining tokens
- Supra3 Pro: 75M parameters, ~150B pretraining tokens
- Supra3 Ultra: 100M parameters, ~200B pretraining tokens
For Supra3 Pro and Ultra, we search for sponsors who give us free access to compute like RTX 5090 32GB or so.
We estimate the total cost of the pro and ultra models at around $600.
For Supra3 Flash Lite and Flash, we do not need sponsors.
If anyone would apply for helping us, we would be really thankful and this person would get early access to new modele, insider information, credit and more!
Contact: here or on discord: lh_tech_ai
We are announcing the Supra3 family with four core SLM models:
- Supra3 Flash Lite: 25M parameters, ~60B pretraining tokens
- Supra3 Flash: 50M parameters, ~100B pretraining tokens
- Supra3 Pro: 75M parameters, ~150B pretraining tokens
- Supra3 Ultra: 100M parameters, ~200B pretraining tokens
For Supra3 Pro and Ultra, we search for sponsors who give us free access to compute like RTX 5090 32GB or so.
We estimate the total cost of the pro and ultra models at around $600.
For Supra3 Flash Lite and Flash, we do not need sponsors.
If anyone would apply for helping us, we would be really thankful and this person would get early access to new modele, insider information, credit and more!
Contact: here or on discord: lh_tech_ai
- 10 replies
reacted to appvoid's post with ๐ฅ 2 days ago
Post
2339
I don't know if it was us or one of you guys or maybe all of us at once but lately we have seen a finetuning/pretraining explosion of models below 200m params and we can't be more happy about it keep coming tinkerers all of this is possible because of you!
- 8 replies
Supra2 100M, BananaMind 2 Pro and GPT-X2.5-135M all releasing at the same time is exciting!
We might going to make Supra2-1B, MAYBE
reacted to Enderchef's post with ๐คโค๏ธ๐๐ฅ 7 days ago
Post
3407
๐ Supra2 100M is out, and multiple other SLM orgs are gaining power!
Following takes a press. Please follow:
fromziro
SupraLabs
AxiomicLabs
Following takes a press. Please follow:
- 8 replies
replied to LH-Tech-AI's post 7 days ago
reacted to LH-Tech-AI's post with ๐คฏ๐๐ฅ 8 days ago
Post
3420
Supra2-100M is out!
Go check it out:
- https://www.reddit.com/r/LocalLLaMA/comments/1velyl9/new_models_supra2100m_base_and_instruct_go_check/
- https://huggingface.co/SupraLabs/Supra2-100M
- SupraLabs/Supra2-100M-Instruct
Give us a like and a follow!!
HAVE FUN ๐ค๐ฅ๐
more coming soon...
Go check it out:
- https://www.reddit.com/r/LocalLLaMA/comments/1velyl9/new_models_supra2100m_base_and_instruct_go_check/
- https://huggingface.co/SupraLabs/Supra2-100M
- SupraLabs/Supra2-100M-Instruct
Give us a like and a follow!!
HAVE FUN ๐ค๐ฅ๐
more coming soon...
- 1 reply
reacted to their post with ๐ฅ 9 days ago
posted an update 9 days ago
reacted to LH-Tech-AI's post with โค๏ธ 9 days ago
Post
2393
Announcing The Supra2 Family And Supra2-100M
Today, we are announcing a brand-new series of SupraLabs models: Supra2
This series will feature various models, including such as:
- ๐ Supra2-Nano (0.4M) โ The smallest Supra2 model.
- ๐ค Supra2-Small (1.4M) โ The tiny model that runs everywhere.
- ๐ช Supra2-Medium (25M) โ Our medium class model in the Supra2 family. The powerful midsizer.
- ๐ฅ Supra2-Pro (100M): base, instruct, reasoning, code, math and more! โ The most capable model yet! A real allrounder for all your everyday tasks.
- ๐จ Supra2-IMG โ our generative text-to-image model
...and many more...
Current progress:
- Nano (0.4M) and Small (1.4M): in training; almost done. Baseline set.
- Medium (25M): coming soon...
- Pro (100M): in training; finishes in 66 hours - Monday, 3rd August 2026, 12:00AM
- IMG: coming soon...
You can support us with a like and follow if you want!
Don't miss our next release! Stay tuned...
Today, we are announcing a brand-new series of SupraLabs models: Supra2
This series will feature various models, including such as:
- ๐ Supra2-Nano (0.4M) โ The smallest Supra2 model.
- ๐ค Supra2-Small (1.4M) โ The tiny model that runs everywhere.
- ๐ช Supra2-Medium (25M) โ Our medium class model in the Supra2 family. The powerful midsizer.
- ๐ฅ Supra2-Pro (100M): base, instruct, reasoning, code, math and more! โ The most capable model yet! A real allrounder for all your everyday tasks.
- ๐จ Supra2-IMG โ our generative text-to-image model
...and many more...
Current progress:
- Nano (0.4M) and Small (1.4M): in training; almost done. Baseline set.
- Medium (25M): coming soon...
- Pro (100M): in training; finishes in 66 hours - Monday, 3rd August 2026, 12:00AM
- IMG: coming soon...
You can support us with a like and follow if you want!
Don't miss our next release! Stay tuned...
- 30 replies
reacted to their post with ๐ง ๐๐ฅ 9 days ago
Post
11066
THIS IS CRAZY! THE MODEL ON THE IMAGE(Supra-50M-Reasoning) answered correctly and its QUANTIZED IN 2BIT! THE RESPONSE IS CORRECT, IN A 15MB SIZE FILE!
- 15 replies
reacted to OppaAI's post with ๐ 9 days ago
Post
1080
After a month of interacting with my AI Waifu, I noticed a few issues in the system; so I decided to spend this week revisiting the systems implemented in Phase 1.0, 1.5 and 2.0, and try to make them to be more like production-grade as much as possible:
1) Memory Degradation - recalled memories are not as good as in the beginning, causing AI Waifu to be more chaotic as she hallucinates over contaminated memories like a bad vicious cycle.
So I transformed the original stateless sqlite-vec vector store to be a simple entity co-mention graph. And even make a studio to visualize the memories stored inside the vector db.
Just by looking at the graph, I saw a couple issues:
a) After 1.5 months of interactions, there should be only one month of pinned memory (in green) over 1.5 months of active memory (in purple). How come pinned memory is in majority over active ones?
I suppose the forgetting curve I had set too aggressive and memory half-life and shelf life too short, active memory got decayed way before monthly consolidation and got lost forever.
b) I saw she memorized me into 3 different entities: my username, my nickname and my Github user ID (leaked into pinned memory, presumbly during nightly dreaming process). 3B small param LLM has hard time to correlation 3 different entities into single person, I may have to harden into one.
2) RAM burst during voice input - for some reason the tensor calculation of SileroVAD of the voice input uses PyTorch, and that's the only place in the whole codebase using torch after removing it from TTS synthesization. By switching to SileroVAD-onnx integrated in the ASR sherpa-onnx, the RAM usage drops at least 0.5GB (after shaving off ~1GB from TTS) by completely remove PyTorch dependencies.
3) Introduced a better Wake Word system using Livekit-Wake word instead of using ASR to do the wake word activation to save computation. Optional features like Speak Verification, Barge-in sensitivity, etc, need to find the optimum settings.
1) Memory Degradation - recalled memories are not as good as in the beginning, causing AI Waifu to be more chaotic as she hallucinates over contaminated memories like a bad vicious cycle.
So I transformed the original stateless sqlite-vec vector store to be a simple entity co-mention graph. And even make a studio to visualize the memories stored inside the vector db.
Just by looking at the graph, I saw a couple issues:
a) After 1.5 months of interactions, there should be only one month of pinned memory (in green) over 1.5 months of active memory (in purple). How come pinned memory is in majority over active ones?
I suppose the forgetting curve I had set too aggressive and memory half-life and shelf life too short, active memory got decayed way before monthly consolidation and got lost forever.
b) I saw she memorized me into 3 different entities: my username, my nickname and my Github user ID (leaked into pinned memory, presumbly during nightly dreaming process). 3B small param LLM has hard time to correlation 3 different entities into single person, I may have to harden into one.
2) RAM burst during voice input - for some reason the tensor calculation of SileroVAD of the voice input uses PyTorch, and that's the only place in the whole codebase using torch after removing it from TTS synthesization. By switching to SileroVAD-onnx integrated in the ASR sherpa-onnx, the RAM usage drops at least 0.5GB (after shaving off ~1GB from TTS) by completely remove PyTorch dependencies.
3) Introduced a better Wake Word system using Livekit-Wake word instead of using ASR to do the wake word activation to save computation. Optional features like Speak Verification, Barge-in sensitivity, etc, need to find the optimum settings.