File size: 1,429 Bytes
57127ef | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 | ---
license: apache-2.0
tags: [chess, reinforcement-learning, grpo]
---
# model_50m_0.23B — RL (GRPO) checkpoints
RL post-training trajectory for the chess pre-to-post compute-allocation study.
The pretraining base and SFT init for this model are
[`model_50m_0.23B`](https://huggingface.co/pavelslab-nyu/Chess-Pretrain-Models/tree/main/model_50m_0.23B) and
[`model_50m_0.23B`](https://huggingface.co/pavelslab-nyu/Chess-SFT-Models/tree/main/model_50m_0.23B).
| | |
|---|---|
| parameters | 50m |
| pretraining tokens | 229,108,466 (0.229B) |
| checkpoints here | 20 (steps 100–2000) |
| checkpoints saved by the run | 40 |
## Steps
`100`, `200`, `300`, `400`, `500`, `600`, `700`, `800`, `900`, `1000`, `1100`, `1200`, `1300`, `1400`, `1500`, `1600`, `1700`, `1800`, `1900`, `2000`
## Loading
Each `global_step_N/` folder is self-contained. The models use a custom
tokenizer (`tokenizer.py`), and the remote-code resolver ignores `subfolder=`,
so download the folder first and load the **local path**:
```python
from huggingface_hub import snapshot_download
from transformers import AutoModelForCausalLM, AutoTokenizer
step = "global_step_2000"
p = snapshot_download("Pre2Post-Chess-RL/Chess-RL-Models", allow_patterns=f"model_50m_0.23B/{step}/*") + f"/model_50m_0.23B/{step}"
model = AutoModelForCausalLM.from_pretrained(p, trust_remote_code=True)
tok = AutoTokenizer.from_pretrained(p, trust_remote_code=True)
```
|