Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
library_name: diffsynth
|
| 4 |
+
tags: [minecraft, world-model, video-generation, inverse-dynamics, agent]
|
| 5 |
+
---
|
| 6 |
+
|
| 7 |
+
# EvoWAM — wild-mine checkpoints
|
| 8 |
+
|
| 9 |
+
Custom checkpoints for the **WAN + IDM closed-loop Minecraft agent** (acts by imagining the next ~1s
|
| 10 |
+
of gameplay with a video model and decoding it into controls, closed-loop on the real game).
|
| 11 |
+
See the GitHub repo `release_wan_idm_wildmine/` for the runnable code (`run_closed_loop.py`).
|
| 12 |
+
|
| 13 |
+
| file | size | what |
|
| 14 |
+
|---|---|---|
|
| 15 |
+
| `wan_fullft/step-52834.safetensors` | 9.4 GB | full fine-tune of the **Wan2.2-TI2V-5B** DiT on Minecraft gameplay + VPT-style action captions |
|
| 16 |
+
| `vpt/4x_idm.model`, `vpt/4x_idm.weights` | ~1.8 GB | **OpenAI VPT** 4x Inverse Dynamics Model (re-hosted for convenience) |
|
| 17 |
+
|
| 18 |
+
## Base models (NOT in this repo — download from source)
|
| 19 |
+
- WAN-2.2-TI2V-5B: https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B
|
| 20 |
+
- Qwen3-VL-8B-Instruct: https://huggingface.co/Qwen/Qwen3-VL-8B-Instruct
|
| 21 |
+
|
| 22 |
+
## Usage
|
| 23 |
+
```
|
| 24 |
+
huggingface-cli download TMarcus/EvoWAM-wildmine-checkpoints --local-dir ./ckpts
|
| 25 |
+
# then point the runner at them:
|
| 26 |
+
# --wan-fullft ckpts/wan_fullft/step-52834.safetensors
|
| 27 |
+
# --idm-model ckpts/vpt/4x_idm.model --idm-weights ckpts/vpt/4x_idm.weights
|
| 28 |
+
```
|
| 29 |
+
|
| 30 |
+
Licenses: the fine-tune derives from Wan2.2-TI2V-5B (see Wan-AI's license); the IDM is OpenAI VPT.
|