Update README.md
Browse files
README.md
CHANGED
|
@@ -6,9 +6,35 @@ tags:
|
|
| 6 |
- community-training
|
| 7 |
metrics:
|
| 8 |
- loss
|
| 9 |
-
---
|
| 10 |
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
-
|
| 14 |
-
Validation Loss: 11.5441
|
|
|
|
| 6 |
- community-training
|
| 7 |
metrics:
|
| 8 |
- loss
|
|
|
|
| 9 |
|
| 10 |
+
HRM-LLM: A truly decentralized, human-like reasoning model built by the community
|
| 11 |
+
|
| 12 |
+
HRM-LLM is a community-driven large language model powered by the Hierarchical Reasoning Model (HRM) architecture. It aims to be truly decentralized: anyone can train, contribute, and scale it forward from anywhere. HRM-LLM is designed to think and work like a human—iterating, refining, and allocating compute adaptively—so it learns efficiently and generalizes across tasks.
|
| 13 |
+
|
| 14 |
+
Why HRM-LLM?
|
| 15 |
+
- Human-like reasoning core: HRM brings hierarchical representations and adaptive computation to mimic iterative human thinking and planning.
|
| 16 |
+
- Adaptive Computation Time (ACT): The model dynamically decides how much “thought” to spend per token—more for hard tokens, less for easy ones.
|
| 17 |
+
- Decentralized and scalable: Anyone can hop in, train a few steps, and push a unified checkpoint to the Hub. Every contribution compounds.
|
| 18 |
+
- Simple, hackable stack: PyTorch + Transformers + Datasets. Easy to extend, easy to improve.
|
| 19 |
+
- Community-aligned progress: Transparent training, open checkpoints, and community governance.
|
| 20 |
+
|
| 21 |
+
What this model aims to do
|
| 22 |
+
- Break down complex problems into stages, reason across them, and refine answers over multiple internal steps.
|
| 23 |
+
- Learn efficient patterns via ACT, saving compute where possible and spending it where it matters most.
|
| 24 |
+
- Become a robust, general-purpose assistant shaped by its global community of contributors.
|
| 25 |
+
|
| 26 |
+
How you can help
|
| 27 |
+
- Train a few steps in Colab (or locally) and push your contribution.
|
| 28 |
+
- Experiment with hyperparameters, tokenizers, datasets, or new HRM blocks.
|
| 29 |
+
- Share insights and logs to improve the next iteration.
|
| 30 |
+
|
| 31 |
+
License
|
| 32 |
+
- This project is licensed under Apache-2.0. You’re free to use, modify, and distribute—with attribution and notice.
|
| 33 |
+
|
| 34 |
+
Jump in and train
|
| 35 |
+
- Colab (1-click): https://colab.research.google.com/drive/1xZNYC-yhwdJxzbpwRekE_rDjTki5CvEv?usp=sharing
|
| 36 |
+
|
| 37 |
+
Quick start: contribute training from your environment
|
| 38 |
+
Run this to join training and push your contribution to the shared checkpoint.
|
| 39 |
|
| 40 |
+
That’s it—share the Colab link, invite contributors, and let the community grow HRM-LLM together.
|
|
|