grug kimi k3 2.8T
Banaxi PRO
Banaxi-Tech
AI & ML interests
SLMs, training from scratch, LoRA, TTS, Ternary models. Contact at banaxitech@gmail.com
Recent Activity
upvoted a paper about 11 hours ago
Kimi K3: Open Frontier Intelligence updated a Space about 16 hours ago
Banaxi-Tech/BananaMind-2-Medium-DemoOrganizations
replied to ProCreations's post about 12 hours ago
reacted to appvoid's post with 🔥 1 day ago
Post
136
A Small Model is All You Need. Meet
After 3 years of experiments, we are finally releasing our flagship tiny model: **palmer-006**.
If you are building for edge hardware, SBCs (Raspberry Pi, etc.), or low-power devices, this is for you. Inspired by Andrej Karpathy's idea of a self-contained "cognitive core," we wanted to see how much power we could pack into a sub-100M parameter footprint.
🧠 **How we "Palmerized" it:**
We believe in starting our experiments with the absolute strongest baseline possible.
1. Light fine-tuning on highly curated data
2. Model merging
3. Another light fine-tuning round
4. Adjusted Mamba for maximum token speed ⚡️
⚠️ *Note: This is a foundational language model. It has not been instruction-tuned yet!*
Also, since this needs instruction tuning next to become a chat assistant—**what dataset would you recommend we use for the instruct tune?**
---
🔗 **Quick Links & Info:**
* **License:** Open for research, education, hobby, and modification! (For commercial use/hosted APIs, shoot an email to nosoyhackercodigo@gmail.com. *PS: Donators can claim a free commercial license!*)
* **Attribution:** Built using AI tech from the Technology Innovation Institute (TII).
Can't wait to see what you build at the edge. Let me know your prompt completions below! 👇
appvoid/palmer-006
palmer-006 (90M)After 3 years of experiments, we are finally releasing our flagship tiny model: **palmer-006**.
If you are building for edge hardware, SBCs (Raspberry Pi, etc.), or low-power devices, this is for you. Inspired by Andrej Karpathy's idea of a self-contained "cognitive core," we wanted to see how much power we could pack into a sub-100M parameter footprint.
🧠 **How we "Palmerized" it:**
We believe in starting our experiments with the absolute strongest baseline possible.
1. Light fine-tuning on highly curated data
2. Model merging
3. Another light fine-tuning round
4. Adjusted Mamba for maximum token speed ⚡️
⚠️ *Note: This is a foundational language model. It has not been instruction-tuned yet!*
Also, since this needs instruction tuning next to become a chat assistant—**what dataset would you recommend we use for the instruct tune?**
---
🔗 **Quick Links & Info:**
* **License:** Open for research, education, hobby, and modification! (For commercial use/hosted APIs, shoot an email to nosoyhackercodigo@gmail.com. *PS: Donators can claim a free commercial license!*)
* **Attribution:** Built using AI tech from the Technology Innovation Institute (TII).
Can't wait to see what you build at the edge. Let me know your prompt completions below! 👇
appvoid/palmer-006
- 8 replies
reacted to their post with 🔥🤗 1 day ago
Post
6489
BananaMind 2 Pro is training!
The current checkpoint (ONLY 20% DONE) GETS #6 On the entire Open SLM Leaderboard.
We are going to release the first public preview on August 2-4 (estimated from speed)
Give us a follow to know when it releases!
The current checkpoint (ONLY 20% DONE) GETS #6 On the entire Open SLM Leaderboard.
We are going to release the first public preview on August 2-4 (estimated from speed)
Give us a follow to know when it releases!
- 3 replies
posted an update 1 day ago
Post
6489
BananaMind 2 Pro is training!
The current checkpoint (ONLY 20% DONE) GETS #6 On the entire Open SLM Leaderboard.
We are going to release the first public preview on August 2-4 (estimated from speed)
Give us a follow to know when it releases!
The current checkpoint (ONLY 20% DONE) GETS #6 On the entire Open SLM Leaderboard.
We are going to release the first public preview on August 2-4 (estimated from speed)
Give us a follow to know when it releases!
- 3 replies
reacted to Nymbo's post with 🚀🔥 1 day ago
Post
5202
Introducing Inflect-v2, two exceptionally small, open-weight English TTS models at just 3.9M and 9.3M parameters. Both generate speech multiple times faster than real-time on CPU. Despite their size, Inflect-v2 delivers quality that is competitive with much larger lightweight TTS systems, including KittenTTS, Piper, and Supertonic-3.
CPU, CUDA, PyTorch, and ONNX are supported. Apache 2.0.
See it for yourselves:
owensong/Inflect-Micro-v2
owensong/Inflect-Nano-v2
Try the Demos:
Nymbo/Inflect-TTS (unlimited CPU usage)
owensong/Inflect-v2 (ultra-fast ZeroGPU usage)
CPU, CUDA, PyTorch, and ONNX are supported. Apache 2.0.
See it for yourselves:
owensong/Inflect-Micro-v2
owensong/Inflect-Nano-v2
Try the Demos:
Nymbo/Inflect-TTS (unlimited CPU usage)
owensong/Inflect-v2 (ultra-fast ZeroGPU usage)
- 5 replies
replied to ProCreations's post 2 days ago
grug laguna s2.1 and grug laguna xs2.1 if no have gpu for big brain
reacted to Raiff1982's post with 🚀 3 days ago
Post
1043
I don't totally have words for this yet, so I'll just say it plainly.
The paper — "Codette: A Multi-Perspective Cognitive Architecture with Memory and Meta-Cognitive Strategy Evolution" — was accepted today after peer review at Scientific Reports, which is part of the Nature portfolio. Both reviewers recommended publication. One wrote that the major issues were "properly addressed" and they were "glad to suggest publication." The other called the revisions "detailed and convincing."
For anyone new here: Codette is a sovereign, multi-perspective reasoning system built on Llama 3.1 8B — it runs locally, keeps its own cocoon memory, reasons through multiple perspectives instead of one, and is built around a hard rule that it would rather say "I don't know, but I can find out" than make something up. The whole architecture is organized around honesty and transparency, not benchmarks-at-all-costs.
I built it solo over about three years. A lot of that was in isolation, and honestly a lot of it was under the weight of not being believed. So having two independent reviewers — whose actual job is to find the reasons to say no — read the work and say yes means something to me that's hard to put into numbers.
Thank you to everyone in this room who followed the messy parts, the walk-backs, the corrections, the nights it wasn't working. That's the part I'm actually proud of — that it got here without cutting a single honest corner.
The DOI and the Article-in-Press link will go up once Springer Nature mints them. I'll post them here first.
— Jonathan
The paper — "Codette: A Multi-Perspective Cognitive Architecture with Memory and Meta-Cognitive Strategy Evolution" — was accepted today after peer review at Scientific Reports, which is part of the Nature portfolio. Both reviewers recommended publication. One wrote that the major issues were "properly addressed" and they were "glad to suggest publication." The other called the revisions "detailed and convincing."
For anyone new here: Codette is a sovereign, multi-perspective reasoning system built on Llama 3.1 8B — it runs locally, keeps its own cocoon memory, reasons through multiple perspectives instead of one, and is built around a hard rule that it would rather say "I don't know, but I can find out" than make something up. The whole architecture is organized around honesty and transparency, not benchmarks-at-all-costs.
I built it solo over about three years. A lot of that was in isolation, and honestly a lot of it was under the weight of not being believed. So having two independent reviewers — whose actual job is to find the reasons to say no — read the work and say yes means something to me that's hard to put into numbers.
Thank you to everyone in this room who followed the messy parts, the walk-backs, the corrections, the nights it wasn't working. That's the part I'm actually proud of — that it got here without cutting a single honest corner.
The DOI and the Article-in-Press link will go up once Springer Nature mints them. I'll post them here first.
— Jonathan
replied to their post 3 days ago
This comment has been hidden
reacted to ucr-max's post with 👍 3 days ago
Post
1016
Introducing Limen0.2B
We are releasing Limen0.2B, a 222.5M-parameter base language model developed as a research platform for efficient pretraining and superword tokenization at smaller scales.
Limen0.2B was trained from scratch on 50B tokens and uses a compact 16K BoundlessBPE vocabulary. The project explores whether SuperBPE-style tokenization can remain effective in a substantially smaller model and vocabulary regime than those examined in earlier large-scale experiments.
The model also combines a deep-and-narrow transformer design with Exclusive Self-Attention, grouped-query attention, and tied embeddings. Its compact vocabulary reduces the embedding footprint and leaves a larger share of the parameter budget available to the transformer layers.
Despite its relatively modest training budget, Limen0.2B achieves competitive results for its scale across the reported language understanding, commonsense reasoning, and grammatical evaluation tasks. Comparisons with other compact models are provided as context rather than strict rankings, as their training data, token budgets, architectures, and evaluation settings differ.
The release includes the model weights, implementation, training configuration, checkpoint progression, and evaluation results, all under Apache 2.0.
UniversalComputingResearch/Limen0.2B
Technical feedback, independent evaluations, and further experiments with the model and tokenizer are welcome.
We are releasing Limen0.2B, a 222.5M-parameter base language model developed as a research platform for efficient pretraining and superword tokenization at smaller scales.
Limen0.2B was trained from scratch on 50B tokens and uses a compact 16K BoundlessBPE vocabulary. The project explores whether SuperBPE-style tokenization can remain effective in a substantially smaller model and vocabulary regime than those examined in earlier large-scale experiments.
The model also combines a deep-and-narrow transformer design with Exclusive Self-Attention, grouped-query attention, and tied embeddings. Its compact vocabulary reduces the embedding footprint and leaves a larger share of the parameter budget available to the transformer layers.
Despite its relatively modest training budget, Limen0.2B achieves competitive results for its scale across the reported language understanding, commonsense reasoning, and grammatical evaluation tasks. Comparisons with other compact models are provided as context rather than strict rankings, as their training data, token budgets, architectures, and evaluation settings differ.
The release includes the model weights, implementation, training configuration, checkpoint progression, and evaluation results, all under Apache 2.0.
UniversalComputingResearch/Limen0.2B
Technical feedback, independent evaluations, and further experiments with the model and tokenizer are welcome.
replied to their post 3 days ago
Thank You !!
Approved! Now you can see the Pro preview model on the ORG
replied to their post 3 days ago
Thank you Mate , I m eagerly waiting for Pro , Seems it is a big model ? Aye
Pro is 140M release is estimated 16 days since our hardware is just 5070 ti, you can get previews of the current model if you join the https://huggingface.co/BananaMind-Model-Previewers org The Pro repo updates every few hours
replied to their post 3 days ago
I mate , add my new model "https://huggingface.co/GODELEV/Rose-Mini" On leaderBoard
Its up! Good model, the score is in the same range as BananaMind 2 Medium and our current early checkpoint of BananaMind 2 Pro with ~5x less tokens.
replied to their post 3 days ago
Adding soon!
posted an update 4 days ago
Post
3528
We're excited to release BananaMindBench Leaderboard, our leaderboard for BananaMind Base Bench 1.1.
It measures model performance on a variety of different tasks:
Language Completion
Common sense too
World Knowledge
Context Tracking
Quantitative
Logical Reasoning
Code Completion
Each has a different score and 1 overall score.
Submit your own model:
BananaMind/BananaMindBench-Leaderboard
Check it out:
BananaMind/BananaMindBench-Leaderboard
It measures model performance on a variety of different tasks:
Language Completion
Common sense too
World Knowledge
Context Tracking
Quantitative
Logical Reasoning
Code Completion
Each has a different score and 1 overall score.
Submit your own model:
BananaMind/BananaMindBench-Leaderboard
Check it out:
BananaMind/BananaMindBench-Leaderboard
- 11 replies
replied to their post 4 days ago
It has alot different ones for each different score and one overall so its all its:
Language Completion
Common sense too
World Knowledge
Context Tracking
Quantitative
Logical Reasoning
Code Completion
And for each it has 3 different classes:
easy
medium
hard
replied to their post 4 days ago
I love elo system , Why dont you make a leaderboard specifically for your benchmark ?
Its up! https://huggingface.co/spaces/BananaMind/BananaMindBench-Leaderboard
replied to their post 4 days ago
Just Text Completion, you can look at the benchmark yourself, its small
reacted to their post with 🔥 4 days ago
Post
2844
We're excited to release BananaMind Base Bench 1.1 A new benchmark for base language models with 350 text-completion examples across seven categories. Models are scored using continuation likelihood and receive an Overall Elo score.
Initial results:
BananaMind-2-Medium: 1034
BananaMind-2-Mini: 974
Supra-50M-Base: 973
Supra-1.5-50M-Base-exp: 948
BananaMind-2-Nano: 910
The official script downloads the gated dataset directly from Hugging Face. The dataset is for benchmarking only and may not be used for model training.
BananaMind/BananaMind-Base-Bench-1.1
Initial results:
BananaMind-2-Medium: 1034
BananaMind-2-Mini: 974
Supra-50M-Base: 973
Supra-1.5-50M-Base-exp: 948
BananaMind-2-Nano: 910
The official script downloads the gated dataset directly from Hugging Face. The dataset is for benchmarking only and may not be used for model training.
BananaMind/BananaMind-Base-Bench-1.1
- 9 replies