Yea prob gonna do that soon
Banaxi PRO
Banaxi-Tech
AI & ML interests
SLMs, training from scratch, LoRA, TTS, Ternary models. AI Interpretability. BCI. Contact at banaxitech@gmail.com
Recent Activity
repliedto their post about 3 hours ago
Introducing BananaMindOS 3.0
- Complete modern UI redesign
- Adds support for Qwen3.5 0.8B, LFM2.5 230M,350M, SmolLM2 360M, Gemma 3 270M.
- Adds Q7,Q6,Q5,Q3,Q1 quantization formats with a easy to use precision slider
- And more!
The new UI includes:
- New 1024×768 High Quality interface.
- Photographic QOI background.
- Transparent BananaMind, CPU, cube, mouse, and Send icons.
- Proper bitmap cursor.
- Rounded translucent panels and cards.
- Modern model-loading progress window.
- Redesigned inference screen with response and prompt panels.
- Localized redraws for the cursor, clicks, loading progress, and precision slider.
Notice: Qwen3.5 0.8B currently generates garbled text, it will be fixed tomorrow.
See it for yourself
Now Available at https://github.com/BananaMind/BananaMindOS
Prebuild ISOs coming soon!
(also press ? + G if you want to load try to load a 6MB RAM model on 5MB may break) reacted to theirpost with 🔥 about 3 hours ago
Introducing BananaMindOS 3.0
- Complete modern UI redesign
- Adds support for Qwen3.5 0.8B, LFM2.5 230M,350M, SmolLM2 360M, Gemma 3 270M.
- Adds Q7,Q6,Q5,Q3,Q1 quantization formats with a easy to use precision slider
- And more!
The new UI includes:
- New 1024×768 High Quality interface.
- Photographic QOI background.
- Transparent BananaMind, CPU, cube, mouse, and Send icons.
- Proper bitmap cursor.
- Rounded translucent panels and cards.
- Modern model-loading progress window.
- Redesigned inference screen with response and prompt panels.
- Localized redraws for the cursor, clicks, loading progress, and precision slider.
Notice: Qwen3.5 0.8B currently generates garbled text, it will be fixed tomorrow.
See it for yourself
Now Available at https://github.com/BananaMind/BananaMindOS
Prebuild ISOs coming soon!
(also press ? + G if you want to load try to load a 6MB RAM model on 5MB may break) reacted to theirpost with 🚀 about 3 hours ago
Introducing BananaMindOS 3.0
- Complete modern UI redesign
- Adds support for Qwen3.5 0.8B, LFM2.5 230M,350M, SmolLM2 360M, Gemma 3 270M.
- Adds Q7,Q6,Q5,Q3,Q1 quantization formats with a easy to use precision slider
- And more!
The new UI includes:
- New 1024×768 High Quality interface.
- Photographic QOI background.
- Transparent BananaMind, CPU, cube, mouse, and Send icons.
- Proper bitmap cursor.
- Rounded translucent panels and cards.
- Modern model-loading progress window.
- Redesigned inference screen with response and prompt panels.
- Localized redraws for the cursor, clicks, loading progress, and precision slider.
Notice: Qwen3.5 0.8B currently generates garbled text, it will be fixed tomorrow.
See it for yourself
Now Available at https://github.com/BananaMind/BananaMindOS
Prebuild ISOs coming soon!
(also press ? + G if you want to load try to load a 6MB RAM model on 5MB may break)Organizations
replied to their post about 3 hours ago
reacted to their post with 🔥🚀 about 3 hours ago
Post
134
Introducing BananaMindOS 3.0
- Complete modern UI redesign
- Adds support for Qwen3.5 0.8B, LFM2.5 230M,350M, SmolLM2 360M, Gemma 3 270M.
- Adds Q7,Q6,Q5,Q3,Q1 quantization formats with a easy to use precision slider
- And more!
The new UI includes:
- New 1024×768 High Quality interface.
- Photographic QOI background.
- Transparent BananaMind, CPU, cube, mouse, and Send icons.
- Proper bitmap cursor.
- Rounded translucent panels and cards.
- Modern model-loading progress window.
- Redesigned inference screen with response and prompt panels.
- Localized redraws for the cursor, clicks, loading progress, and precision slider.
Notice: Qwen3.5 0.8B currently generates garbled text, it will be fixed tomorrow.
See it for yourself
Now Available at https://github.com/BananaMind/BananaMindOS
Prebuild ISOs coming soon!
(also press ? + G if you want to load try to load a 6MB RAM model on 5MB may break)
- Complete modern UI redesign
- Adds support for Qwen3.5 0.8B, LFM2.5 230M,350M, SmolLM2 360M, Gemma 3 270M.
- Adds Q7,Q6,Q5,Q3,Q1 quantization formats with a easy to use precision slider
- And more!
The new UI includes:
- New 1024×768 High Quality interface.
- Photographic QOI background.
- Transparent BananaMind, CPU, cube, mouse, and Send icons.
- Proper bitmap cursor.
- Rounded translucent panels and cards.
- Modern model-loading progress window.
- Redesigned inference screen with response and prompt panels.
- Localized redraws for the cursor, clicks, loading progress, and precision slider.
Notice: Qwen3.5 0.8B currently generates garbled text, it will be fixed tomorrow.
See it for yourself
Now Available at https://github.com/BananaMind/BananaMindOS
Prebuild ISOs coming soon!
(also press ? + G if you want to load try to load a 6MB RAM model on 5MB may break)
- 3 replies
Its released
AI waifu...
posted an update about 3 hours ago
Post
134
Introducing BananaMindOS 3.0
- Complete modern UI redesign
- Adds support for Qwen3.5 0.8B, LFM2.5 230M,350M, SmolLM2 360M, Gemma 3 270M.
- Adds Q7,Q6,Q5,Q3,Q1 quantization formats with a easy to use precision slider
- And more!
The new UI includes:
- New 1024×768 High Quality interface.
- Photographic QOI background.
- Transparent BananaMind, CPU, cube, mouse, and Send icons.
- Proper bitmap cursor.
- Rounded translucent panels and cards.
- Modern model-loading progress window.
- Redesigned inference screen with response and prompt panels.
- Localized redraws for the cursor, clicks, loading progress, and precision slider.
Notice: Qwen3.5 0.8B currently generates garbled text, it will be fixed tomorrow.
See it for yourself
Now Available at https://github.com/BananaMind/BananaMindOS
Prebuild ISOs coming soon!
(also press ? + G if you want to load try to load a 6MB RAM model on 5MB may break)
- Complete modern UI redesign
- Adds support for Qwen3.5 0.8B, LFM2.5 230M,350M, SmolLM2 360M, Gemma 3 270M.
- Adds Q7,Q6,Q5,Q3,Q1 quantization formats with a easy to use precision slider
- And more!
The new UI includes:
- New 1024×768 High Quality interface.
- Photographic QOI background.
- Transparent BananaMind, CPU, cube, mouse, and Send icons.
- Proper bitmap cursor.
- Rounded translucent panels and cards.
- Modern model-loading progress window.
- Redesigned inference screen with response and prompt panels.
- Localized redraws for the cursor, clicks, loading progress, and precision slider.
Notice: Qwen3.5 0.8B currently generates garbled text, it will be fixed tomorrow.
See it for yourself
Now Available at https://github.com/BananaMind/BananaMindOS
Prebuild ISOs coming soon!
(also press ? + G if you want to load try to load a 6MB RAM model on 5MB may break)
- 3 replies
Late night release today! In 20 minutes we will get BananaMindOS 3.0
reacted to OppaAI's post with 😔 about 5 hours ago
Post
2576
Congrat to HuggingFace Team
https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/
https://techcrunch.com/2026/08/26/nvidia-closes-in-on-hugging-face-acquisition/
- 1 reply
replied to KlondikeDev's post about 12 hours ago
you should invite @appvoid
reacted to their post with 👀 about 12 hours ago
Post
1473
We're delaying BananaMind 2.1!
When BananaMind 2.1 Lite was almost done, we benchmarked it and the results we're worse than BananaMind 2 Mini.
We're going to spend alot more time in research on tiny models and then scaling up our techniques to the actual BananaMind 2.1 models!
We're also announcing these new models:
BananaMind 2.1 Coder: A 149M instruction tuned coder model trained on 75B tokens + 10B tokens of stack-v3-train.
BananaMind 2.1 Pico: A 1M parameter model trained on 22B tokens of data.
We also may release BananaMind 2.1 Large with around 100M parameters depending on how much compute we have.
Please give us a follow!
BananaMind
@Banaxi-Tech
---
@vovaRL
@DedeProGames
When BananaMind 2.1 Lite was almost done, we benchmarked it and the results we're worse than BananaMind 2 Mini.
We're going to spend alot more time in research on tiny models and then scaling up our techniques to the actual BananaMind 2.1 models!
We're also announcing these new models:
BananaMind 2.1 Coder: A 149M instruction tuned coder model trained on 75B tokens + 10B tokens of stack-v3-train.
BananaMind 2.1 Pico: A 1M parameter model trained on 22B tokens of data.
We also may release BananaMind 2.1 Large with around 100M parameters depending on how much compute we have.
Please give us a follow!
@Banaxi-Tech
---
@vovaRL
@DedeProGames
- 7 replies
@Banaxi-Tech nice i joined. whats it for?
https://huggingface.co/bananamind-research-community/bananamind-2.0.1 will be updated for new experiments
idk we will post new arch there before release and other can share their own custom arch too
@Banaxi-Tech excited to see what Bananamind 2.1 brings! i trust it will be very novel lol
if you want previous go joinhttps://huggingface.co/bananamind-research-community
Im doing loops now too lol
I've made an architecture that is arguably better than loops but since there doing so well i might just add them to my arch combining both mechanisms
Do you mean the Qana mini arch?
Also we're now using n-gram embeddings from Qwen Next
That's just called Engram Conditional Memory; it was introduced by DeepSeek, not Qwen.
https://huggingface.co/Qwen/Qwen3.8-Flash-Next
maybe it was but i first saw it here
You should only use some loops, like lets say you have 4 layers then:
L1 🠮 L2 🠮 L2 🠮 L3 🠮 L4 so 5 effective layers not
L1 🠮 L1 🠮 L2 🠮 L2 🠮 L3 🠮 L3 🠮 L4 🠮 L4 because our tests on 1M:
| Metric | All looped (6 blocks) | Partial looped (4 blocks) | No loop (3 blocks) |
|---|---|---|---|
| Base Bench Elo | 875 | 885 | 884 |
| Base raw accuracy | 33.14% | 34.57% | 33.71% |
| Base weighted accuracy | 32.54% | 33.65% | 33.57% |
| ARC Easy acc_norm | 30.98% | 29.50% | 30.35% |
| ARC Challenge acc_norm | 21.33% | 22.53% | 22.10% |
| PIQA acc_norm | 52.45% | 54.30% | 52.56% |
| HellaSwag acc_norm | 27.28% | 27.04% | 26.93% |
| ArithMark-3 acc_norm | 30.40% | 33.00% | 30.80% |
| INT Index | 3.88 | 5.37 | 3.93 |
| Training throughput | 344K tok/s | 492K tok/s | 553K tok/s |
Also we're now using n-gram embeddings from Qwen Next
Very early though we're not going to do loops only probrably 3-6 tower + loops + one new architecture surprise
We also not sure yet if we will use but right now in our testing we got these results:
Model: 1M parameter model on 2B tokens of streamed FW EDU
| Metric | All looped (6 blocks) | Partial looped (4 blocks) | No loop (3 blocks) |
|---|---|---|---|
| Base Bench Elo | 875 | 885 | 884 |
| Base raw accuracy | 33.14% | 34.57% | 33.71% |
| Base weighted accuracy | 32.54% | 33.65% | 33.57% |
| ARC Easy acc_norm | 30.98% | 29.50% | 30.35% |
| ARC Challenge acc_norm | 21.33% | 22.53% | 22.10% |
| PIQA acc_norm | 52.45% | 54.30% | 52.56% |
| HellaSwag acc_norm | 27.28% | 27.04% | 26.93% |
| ArithMark-3 acc_norm | 30.40% | 33.00% | 30.80% |
| INT Index | 3.88 | 5.37 | 3.93 |
| Training throughput | 344K tok/s | 492K tok/s | 553K tok/s |