Add support for decision transformer (#795)

* Add support for decision transformer (Closes #794)

* Comment out supported decision transformer models

Models are in the `onnx-community` org on HF

README.md

@@ -250,7 +250,7 @@ You can refine your search by selecting the task you're interested in (e.g., [te
 
 | Task                     | ID | Description | Supported? |
 |--------------------------|----|-------------|------------|
-| [Reinforcement Learning](https://huggingface.co/tasks/reinforcement-learning)   |  n/a  | Learning from actions by interacting with an environment through trial and error and receiving rewards (negative or positive) as feedback. | ❌ |
+| [Reinforcement Learning](https://huggingface.co/tasks/reinforcement-learning)   |  n/a  | Learning from actions by interacting with an environment through trial and error and receiving rewards (negative or positive) as feedback. | ✅ |
 
 
 
@@ -276,6 +276,7 @@ You can refine your search by selecting the task you're interested in (e.g., [te
 1. **[ConvNeXTV2](https://huggingface.co/docs/transformers/model_doc/convnextv2)** (from Facebook AI) released with the paper [ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders](https://arxiv.org/abs/2301.00808) by Sanghyun Woo, Shoubhik Debnath, Ronghang Hu, Xinlei Chen, Zhuang Liu, In So Kweon, Saining Xie.
 1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
 1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
+1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
 1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
 1. **[Depth Anything](https://huggingface.co/docs/transformers/main/model_doc/depth_anything)** (from University of Hong Kong and TikTok) released with the paper [Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data](https://arxiv.org/abs/2401.10891) by Lihe Yang, Bingyi Kang, Zilong Huang, Xiaogang Xu, Jiashi Feng, Hengshuang Zhao.
 1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.

docs/snippets/5_supported-tasks.snippet

@@ -67,4 +67,4 @@
 
 | Task                     | ID | Description | Supported? |
 |--------------------------|----|-------------|------------|
-| [Reinforcement Learning](https://huggingface.co/tasks/reinforcement-learning)   |  n/a  | Learning from actions by interacting with an environment through trial and error and receiving rewards (negative or positive) as feedback. | ❌ |
+| [Reinforcement Learning](https://huggingface.co/tasks/reinforcement-learning)   |  n/a  | Learning from actions by interacting with an environment through trial and error and receiving rewards (negative or positive) as feedback. | ✅ |

docs/snippets/6_supported-models.snippet

@@ -21,6 +21,7 @@
 1. **[ConvNeXTV2](https://huggingface.co/docs/transformers/model_doc/convnextv2)** (from Facebook AI) released with the paper [ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders](https://arxiv.org/abs/2301.00808) by Sanghyun Woo, Shoubhik Debnath, Ronghang Hu, Xinlei Chen, Zhuang Liu, In So Kweon, Saining Xie.
 1. **[DeBERTa](https://huggingface.co/docs/transformers/model_doc/deberta)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
 1. **[DeBERTa-v2](https://huggingface.co/docs/transformers/model_doc/deberta-v2)** (from Microsoft) released with the paper [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://arxiv.org/abs/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen.
+1. **[Decision Transformer](https://huggingface.co/docs/transformers/model_doc/decision_transformer)** (from Berkeley/Facebook/Google) released with the paper [Decision Transformer: Reinforcement Learning via Sequence Modeling](https://arxiv.org/abs/2106.01345) by Lili Chen, Kevin Lu, Aravind Rajeswaran, Kimin Lee, Aditya Grover, Michael Laskin, Pieter Abbeel, Aravind Srinivas, Igor Mordatch.
 1. **[DeiT](https://huggingface.co/docs/transformers/model_doc/deit)** (from Facebook) released with the paper [Training data-efficient image transformers & distillation through attention](https://arxiv.org/abs/2012.12877) by Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou.
 1. **[Depth Anything](https://huggingface.co/docs/transformers/main/model_doc/depth_anything)** (from University of Hong Kong and TikTok) released with the paper [Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data](https://arxiv.org/abs/2401.10891) by Lihe Yang, Bingyi Kang, Zilong Huang, Xiaogang Xu, Jiashi Feng, Hengshuang Zhao.
 1. **[DETR](https://huggingface.co/docs/transformers/model_doc/detr)** (from Facebook) released with the paper [End-to-End Object Detection with Transformers](https://arxiv.org/abs/2005.12872) by Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, Sergey Zagoruyko.

scripts/supported_models.py

@@ -299,6 +299,22 @@ SUPPORTED_MODELS = {
             'sileod/deberta-v3-large-tasksource-nli',
         ],
     },
+    # TODO: Add back in v3
+    # 'decision-transformer': {
+    #     # Reinforcement learning
+    #     'reinforcement-learning': [
+    #         'edbeeching/decision-transformer-gym-hopper-expert',
+    #         'edbeeching/decision-transformer-gym-hopper-medium',
+    #         'edbeeching/decision-transformer-gym-hopper-medium-replay',
+    #         'edbeeching/decision-transformer-gym-hopper-expert-new',
+    #         'edbeeching/decision-transformer-gym-halfcheetah-expert',
+    #         'edbeeching/decision-transformer-gym-halfcheetah-medium',
+    #         'edbeeching/decision-transformer-gym-halfcheetah-medium-replay',
+    #         'edbeeching/decision-transformer-gym-walker2d-expert',
+    #         'edbeeching/decision-transformer-gym-walker2d-medium',
+    #         'edbeeching/decision-transformer-gym-walker2d-medium-replay',
+    #     ],
+    # },
     'deit': {
         # Image classification
         'image-classification': [

src/models.js

@@ -5481,6 +5481,17 @@ export class EfficientNetForImageClassification extends EfficientNetPreTrainedMo
 }
 //////////////////////////////////////////////////
 
+//////////////////////////////////////////////////
+// Decision Transformer models
+export class DecisionTransformerPreTrainedModel extends PreTrainedModel { }
+
+/**
+ * The model builds upon the GPT2 architecture to perform autoregressive prediction of actions in an offline RL setting.
+ * Refer to the paper for more details: https://arxiv.org/abs/2106.01345
+ */
+export class DecisionTransformerModel extends DecisionTransformerPreTrainedModel { }
+
+//////////////////////////////////////////////////
 
 //////////////////////////////////////////////////
 // AutoModels, used to simplify construction of PreTrainedModels
@@ -5607,6 +5618,7 @@ const MODEL_MAPPING_NAMES_ENCODER_ONLY = new Map([
     ['hifigan', ['SpeechT5HifiGan', SpeechT5HifiGan]],
     ['efficientnet', ['EfficientNetModel', EfficientNetModel]],
 
+    ['decision_transformer', ['DecisionTransformerModel', DecisionTransformerModel]],
 ]);
 
 const MODEL_MAPPING_NAMES_ENCODER_DECODER = new Map([