MinimaLabs/min-spark-1.1
Text Generation • 5.76M • Updated • 223 • 12
🤝 Open to Collab
LH-Tech AI
LH-Tech-AI
AI & ML interests
Small AI and ML models. Trained by myself. Completely OpenSource. For you. | Reddit: https://www.reddit.com/user/LH-Tech_AI/
Recent Activity
new activity 1 day ago
Qyrou/chat-titles-filtered-115K:Rebrand: replace stale SupraLabs references with Qyrou liked a model 1 day ago
MinimaLabs/min-spark-1.1Organizations
Rebrand: replace stale SupraLabs references with Qyrou
2
#2 opened 2 days ago
by 
QyrouNnet-AI
liked a
model 1 day ago
reacted to Bc-AI's post with 🔥 1 day ago
Post
3499
New update! We are currently training a few new models now! Our 3rd generation main LLM standard edition is in training right now. We are also training a new LLM line called Tiny Coder around 350~ish M params. Thanks to @Banaxi-Tech for inspiring the architecture with his Bananamind-2.1-unified test model. Thanks to our beta testers: @juiceb0xc0de @ProCreations @Sbui503 @Fishtiks @MUK-IS-GOAT
- 1 reply
liked a
model 1 day ago
reacted to Banaxi-Tech's post with 🔥 1 day ago
Post
3702
We're releasing BananaMind 2.1 Unified, a 35M three-tower relay model where the two output towers can only talk to each other through a silent middle tower that has no output head and no loss term.
Tower B trains entirely on indirect gradient. It was never told what to predict. It woke up anyway. Jacobian lens shows it carrying the correct answer ("Paris", "oxygen", "blue") at its deepest layer. Its bridge gates grew 4-47x from init. Feed it from only one side and the representations collapse to junk — it needs both outer towers to become semantic.
The PIQA result is the cleanest demonstration: Tower A alone scores 50.11 (chance). Tower C alone 52.12. Full system 61.75. All physical reasoning lives in the integration. The 35M three-tower beats the 50M single-tower BananaMind 2 Medium on PIQA.
Trained on 38B tokens in ~7.5 hours on 8x RTX PRO 6000. Ships with 7 ablation modes so you can surgically cut the model apart without retraining. Full training logs, J-lens fits, and eval outputs for every mode included.
This is a research model. It is interesting because you can take it apart.
BananaMind/BananaMind-2.1-Unified
Follow us for more:
BananaMind
@Banaxi-Tech
Tower B trains entirely on indirect gradient. It was never told what to predict. It woke up anyway. Jacobian lens shows it carrying the correct answer ("Paris", "oxygen", "blue") at its deepest layer. Its bridge gates grew 4-47x from init. Feed it from only one side and the representations collapse to junk — it needs both outer towers to become semantic.
The PIQA result is the cleanest demonstration: Tower A alone scores 50.11 (chance). Tower C alone 52.12. Full system 61.75. All physical reasoning lives in the integration. The 35M three-tower beats the 50M single-tower BananaMind 2 Medium on PIQA.
Trained on 38B tokens in ~7.5 hours on 8x RTX PRO 6000. Ships with 7 ablation modes so you can surgically cut the model apart without retraining. Full training logs, J-lens fits, and eval outputs for every mode included.
This is a research model. It is interesting because you can take it apart.
BananaMind/BananaMind-2.1-Unified
Follow us for more:
@Banaxi-Tech
- 11 replies
replied to Banaxi-Tech's post 1 day ago
Looks cool! Can't wait to test these!
But we will beat this with our Supra3 family soon (coming in around 4 to 8 weeks)! 🔥🙃
Stay tuned! :D
reacted to Banaxi-Tech's post with 🔥 1 day ago
Post
3708
We're announcing our BananaMind 2.1 model series!
The models will include:
- BananaMind 2.1 Nano: 10M parameters with 60B tokens.
- BananaMind 2.1 Lite: 25M parameters with 40B tokens.
- BananaMind 2.1 Flash: 50M parameters with 55B tokens.
- BananaMind 2.1 Pro: 135M-145M parameters (still deciding) with 100B tokens.
These model will use a multi tower architecture (like BananaMind/BananaMind-2.1-Unified) with some more architectural changes.
BananaMind 2.1 Pro will probrably use 2 no output towers, instead of one!
We're currently training some experimental models based on this architecture to see its scaling!
Follow us:
BananaMind
@Banaxi-Tech
@vovaRL
@DedeProGames
bananamind-research-community
The models will include:
- BananaMind 2.1 Nano: 10M parameters with 60B tokens.
- BananaMind 2.1 Lite: 25M parameters with 40B tokens.
- BananaMind 2.1 Flash: 50M parameters with 55B tokens.
- BananaMind 2.1 Pro: 135M-145M parameters (still deciding) with 100B tokens.
These model will use a multi tower architecture (like BananaMind/BananaMind-2.1-Unified) with some more architectural changes.
BananaMind 2.1 Pro will probrably use 2 no output towers, instead of one!
We're currently training some experimental models based on this architecture to see its scaling!
Follow us:
@Banaxi-Tech
@vovaRL
@DedeProGames
- 10 replies
updated 2
models 1 day ago
liked 3
models 2 days ago
Performance compared to Supra 1.5 50m exp
1
#2 opened 2 days ago
by 
Dalek70
liked 2
models 3 days ago
supra u are no uploading new models ur so slow
🤗 1
4
#1 opened 3 days ago
by 
RexTRO111
liked a
model 3 days ago
published a
model 3 days ago
updated a
collection 3 days ago