CrafterDojo v1.0 weights (safetensors)

MANIFEST.json

@@ -0,0 +1,59 @@
+{
+  "vpt/tiny.safetensors": {
+    "source": "vpt_bc/sv_gh_pp-10B_20ep/tiny/last",
+    "tensors": 100,
+    "sha256": "2e9b8d2c3fe0a13bc0927d29d5eb716e97b26a8b71db68588b6d2c3d0dff9944",
+    "bytes": 12940452
+  },
+  "vpt/base.safetensors": {
+    "source": "vpt_bc/sv_gh_pp-10B_20ep/base/last",
+    "tensors": 132,
+    "sha256": "dd7ddaa128bc3b2adb71002329410a8b6bd17ea3f5c9a87c74593dea7c7976b5",
+    "bytes": 63455452
+  },
+  "vpt/large.safetensors": {
+    "source": "vpt_bc/sv_gh_pp-10B_20ep/large/last",
+    "tensors": 132,
+    "sha256": "199985cb18b77613e04f5bab17d2c94ed679f8a6eaff64482deb634f6d751027",
+    "bytes": 257128276
+  },
+  "steve1/steve1.safetensors": {
+    "source": "csteve1/sv_gh_pp-10B_20ep/swsvgh_rep39_last/last",
+    "tensors": 166,
+    "sha256": "a09ce91327f0ff8e0b3340ab1714537980508078ae7426950dc8db156fd0974b",
+    "bytes": 65438812
+  },
+  "steve1/steve1_prior.safetensors": {
+    "source": "steve1/cvae_prior_1k/steve1_prior_sv_gh_pp-10B_20ep_reph39",
+    "tensors": 20,
+    "sha256": "548d7d7fb8c6dac6ad2637bd12d391b594b81be3f47efba94f0ccdb5e91398f6",
+    "bytes": 9469768
+  },
+  "crafterclip/crafterclip.safetensors": {
+    "source": "crafterclip/sv_gh_pp-10B_20ep_minbal/rep39/best",
+    "tensors": 334,
+    "aliases": 454,
+    "sha256": "2fba12ed345a26bf37a6e99b6509e8a721085d1dc809f461568de23ff0a99357",
+    "bytes": 633997584
+  },
+  "vpt/tiny.yaml": {
+    "source": "vpt/tiny.model",
+    "sha256": "98eadc0b1ce5a96096ec57e239b6bf5171c9ad2dd325fc1c930aa0d21cd07c0e",
+    "bytes": 865
+  },
+  "vpt/base.yaml": {
+    "source": "vpt/base.model",
+    "sha256": "189e36b6f952a5cf3ef1443e7ac631baccb1e40f1e5f7bf0be83effb8ce793c7",
+    "bytes": 865
+  },
+  "vpt/large.yaml": {
+    "source": "vpt/large.model",
+    "sha256": "4a5590eaf206f0adc8f3a8dd6150e592fe16173ff70a0f5a9a51cbf37374d546",
+    "bytes": 866
+  },
+  "vpt/base_lora.yaml": {
+    "source": "vpt/base_lora.model",
+    "sha256": "d268401bc79e7746e2a9d78d0bdf7d732ce8f0f28c80c9a479e91d73d097b42d",
+    "bytes": 903
+  }
+}

README.md

@@ -0,0 +1,49 @@
+---
+license: mit
+tags:
+  - reinforcement-learning
+  - embodied-ai
+  - behavior-cloning
+  - crafter
+  - craftax
+  - video-language
+---
+
+# CrafterDojo v1 weights
+
+Pretrained models from **CrafterDojo: An Integrated Research Suite for Building Open-Ended Embodied Agents in Crafter**
+([paper](https://arxiv.org/abs/2508.13530), [code](https://github.com/frechele/CrafterDojo), [project](https://sites.google.com/view/crafterdojo)).
+
+| File | Model | Notes |
+|---|---|---|
+| `vpt/tiny.safetensors` | CrafterVPT-tiny | behavior cloning, last checkpoint |
+| `vpt/base.safetensors` | CrafterVPT-base | behavior cloning, last checkpoint |
+| `vpt/large.safetensors` | CrafterVPT-large | behavior cloning, last checkpoint |
+| `crafterclip/crafterclip.safetensors` | CrafterCLIP | video–caption encoder (MineCLIP architecture, ViT-B/16 backbone) |
+| `steve1/steve1.safetensors` | CrafterSteve-1 | instruction-following policy (LoRA on CrafterVPT-base) |
+| `steve1/steve1_prior.safetensors` | CrafterSteve-1 prior | text → visual-goal CVAE prior |
+| `vpt/*.yaml` | architecture configs | same files as `models/vpt/*.yaml` in the code repository |
+
+All models are trained on the `sv_gh_pp-10B_20ep` CrafterPlay mixture (Survival + Gatherer experts) in Craftax-Classic.
+Each `.safetensors` file records its training run in the header metadata.
+
+## Usage
+
+```bash
+git clone https://github.com/frechele/CrafterDojo && cd CrafterDojo
+bash scripts/download_weights.sh      # downloads this repository (revision v1.0) into models/
+```
+
+Load with the repository's helpers (`crafterdojo.common.checkpoint.load_state_dict`), which also restore the
+shared CLIP tensors in CrafterCLIP (stored once, listed in the `aliases` metadata entry).
+
+## Citation
+
+```bibtex
+@article{park2025crafterdojo,
+  title   = {CrafterDojo: An Integrated Research Suite for Building Open-Ended Embodied Agents in Crafter},
+  author  = {Park, Junyeong and Cho, Hyeonseo and Ahn, Sungjin},
+  journal = {arXiv preprint arXiv:2508.13530},
+  year    = {2025}
+}
+```

crafterclip/crafterclip.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fba12ed345a26bf37a6e99b6509e8a721085d1dc809f461568de23ff0a99357
+size 633997584

steve1/steve1.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a09ce91327f0ff8e0b3340ab1714537980508078ae7426950dc8db156fd0974b
+size 65438812

steve1/steve1_prior.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:548d7d7fb8c6dac6ad2637bd12d391b594b81be3f47efba94f0ccdb5e91398f6
+size 9469768

vpt/base.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd7ddaa128bc3b2adb71002329410a8b6bd17ea3f5c9a87c74593dea7c7976b5
+size 63455452

vpt/base.yaml

@@ -0,0 +1,36 @@
+model:
+  args:
+    net:
+      args:
+        active_reward_monitors:
+          env_reward: 1
+        attention_heads: 4
+        attention_mask_style: clipped_causal
+        attention_memory_size: 128
+        diff_mlp_embedding: false
+        hidsize: 512
+        img_shape:
+        - 144
+        - 144
+        - 3
+        impala_chans:
+        - 16
+        - 32
+        - 32
+        impala_kwargs:
+          post_pool_groups: 1
+        impala_width: 1
+        init_norm_kwargs:
+          batch_norm: false
+          group_norm_groups: 1
+        n_recurrence_layers: 4
+        only_img_input: true
+        pointwise_ratio: 4
+        pointwise_use_activation: false
+        recurrence_is_residual: true
+        recurrence_type: transformer
+        timesteps: 64
+        use_pointwise_layer: true
+        use_pre_lstm_ln: false
+    pi_head_opts:
+      temperature: 1.0

vpt/base_lora.yaml

@@ -0,0 +1,39 @@
+model:
+  args:
+    net:
+      args:
+        active_reward_monitors:
+          env_reward: 1
+        attention_heads: 4
+        attention_mask_style: clipped_causal
+        attention_memory_size: 128
+        diff_mlp_embedding: false
+        hidsize: 512
+        img_shape:
+        - 144
+        - 144
+        - 3
+        impala_chans:
+        - 16
+        - 32
+        - 32
+        impala_kwargs:
+          post_pool_groups: 1
+        impala_width: 1
+        init_norm_kwargs:
+          batch_norm: false
+          group_norm_groups: 1
+        n_recurrence_layers: 4
+        only_img_input: true
+        pointwise_ratio: 4
+        pointwise_use_activation: false
+        recurrence_is_residual: true
+        recurrence_type: transformer
+        timesteps: 64
+        use_pointwise_layer: true
+        use_pre_lstm_ln: false
+    pi_head_opts:
+      temperature: 1.0
+    lora:
+      alpha: 16
+      r: 16

vpt/large.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:199985cb18b77613e04f5bab17d2c94ed679f8a6eaff64482deb634f6d751027
+size 257128276

vpt/large.yaml

@@ -0,0 +1,36 @@
+model:
+  args:
+    net:
+      args:
+        active_reward_monitors:
+          env_reward: 1
+        attention_heads: 8
+        attention_mask_style: clipped_causal
+        attention_memory_size: 128
+        diff_mlp_embedding: false
+        hidsize: 1024
+        img_shape:
+        - 144
+        - 144
+        - 3
+        impala_chans:
+        - 16
+        - 32
+        - 32
+        impala_kwargs:
+          post_pool_groups: 1
+        impala_width: 4
+        init_norm_kwargs:
+          batch_norm: false
+          group_norm_groups: 1
+        n_recurrence_layers: 4
+        only_img_input: true
+        pointwise_ratio: 4
+        pointwise_use_activation: false
+        recurrence_is_residual: true
+        recurrence_type: transformer
+        timesteps: 64
+        use_pointwise_layer: true
+        use_pre_lstm_ln: false
+    pi_head_opts:
+      temperature: 1.0

vpt/tiny.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e9b8d2c3fe0a13bc0927d29d5eb716e97b26a8b71db68588b6d2c3d0dff9944
+size 12940452

vpt/tiny.yaml

@@ -0,0 +1,36 @@
+model:
+  args:
+    net:
+      args:
+        active_reward_monitors:
+          env_reward: 1
+        attention_heads: 4
+        attention_mask_style: clipped_causal
+        attention_memory_size: 128
+        diff_mlp_embedding: false
+        hidsize: 128
+        img_shape:
+        - 144
+        - 144
+        - 3
+        impala_chans:
+        - 16
+        - 32
+        - 32
+        impala_kwargs:
+          post_pool_groups: 1
+        impala_width: 1
+        init_norm_kwargs:
+          batch_norm: false
+          group_norm_groups: 1
+        n_recurrence_layers: 2
+        only_img_input: true
+        pointwise_ratio: 4
+        pointwise_use_activation: false
+        recurrence_is_residual: true
+        recurrence_type: transformer
+        timesteps: 64
+        use_pointwise_layer: true
+        use_pre_lstm_ln: false
+    pi_head_opts:
+      temperature: 1.0