unsloth/Qwen3.8-27B-GGUF
27B • Updated • 5.13M • 2.37k
alkinun
AtAndDev
AI & ML interests
decentralize
Recent Activity
liked a model about 6 hours ago
unsloth/Qwen3.8-27B-GGUF liked a dataset about 13 hours ago
openbmb/Ultra-FineWeb-L1 liked a dataset about 13 hours ago
Anthropic/claude-protein-binder-designOrganizations
liked a
model about 6 hours ago
liked 2
datasets about 13 hours ago
liked 2
models about 13 hours ago
reacted to Banaxi-Tech's post with 🚀👀🔥 about 15 hours ago
Post
2335
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
- 4 replies
reacted to Felladrin's post with 👀🔥👍 about 16 hours ago
Post
4021
I've open-sourced the trainer I've been using to build tiny language models from scratch, together with the 95M base model I trained with it.
The trainer runs on Deno (https://deno.com, cross-platform), trains on WebGPU, and it writes GGUF directly. No Python/PyTorch. The weights live in a GGUF file from the first step to the last, so every checkpoint is already something llama.cpp can load.
The model is Felladrin/Minueza-3-95M-Base: 94.7M parameters, 1.95B tokens seen, 8192 context.
And here’s the repository on GitHub: https://github.com/felladrin/gguf-trainer
Here on Hugging Face, I published the optimizer state next to the weights, so you can continue the pretraining instead of starting over.
Or start your own from nothing:
And the docs are written for coding agents, so you can point your agent of choice at the GitHub repo and have it drive the whole pipeline.
The trainer runs on Deno (https://deno.com, cross-platform), trains on WebGPU, and it writes GGUF directly. No Python/PyTorch. The weights live in a GGUF file from the first step to the last, so every checkpoint is already something llama.cpp can load.
The model is Felladrin/Minueza-3-95M-Base: 94.7M parameters, 1.95B tokens seen, 8192 context.
And here’s the repository on GitHub: https://github.com/felladrin/gguf-trainer
Here on Hugging Face, I published the optimizer state next to the weights, so you can continue the pretraining instead of starting over.
Or start your own from nothing:
deno run -A cli.ts demo trains a tiny one end to end in under a minute.And the docs are written for coding agents, so you can point your agent of choice at the GitHub repo and have it drive the whole pipeline.
- 36 replies
liked a
model about 16 hours ago
ill be on this list soon!
hopefully
reacted to Enderchef's post with 👀🚀🤗🔥 about 19 hours ago
Post
1817
Do you support the SLM(1M-150M parameter) community?
If so, join the SLM discord(https://discord.gg/BBYaERvvn), and give these orgs some follows!
BananaMind
SupraLabs
fromziro
AxiomicLabs
If so, join the SLM discord(https://discord.gg/BBYaERvvn), and give these orgs some follows!
- 1 reply
liked 2
models 1 day ago
liked a
dataset 1 day ago