Upload . with huggingface_hub

.gitattributes

@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+replay.mp4 filter=lfs diff=lfs merge=lfs -text

.summary/0/events.out.tfevents.1678544028.qgallouedec-MS-7C84

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72daab09d369100793f8ed65f3e5d014baa59a347fe88129c849dc254ac5f8b8
+size 4494124

README.md

@@ -0,0 +1,56 @@
+---
+library_name: sample-factory
+tags:
+- deep-reinforcement-learning
+- reinforcement-learning
+- sample-factory
+model-index:
+- name: APPO
+  results:
+  - task:
+      type: reinforcement-learning
+      name: reinforcement-learning
+    dataset:
+      name: stick-pull-v2
+      type: stick-pull-v2
+    metrics:
+    - type: mean_reward
+      value: 55.63 +/- 118.71
+      name: mean_reward
+      verified: false
+---
+
+A(n) **APPO** model trained on the **stick-pull-v2** environment.
+
+This model was trained using Sample-Factory 2.0: https://github.com/alex-petrenko/sample-factory.
+Documentation for how to use Sample-Factory can be found at https://www.samplefactory.dev/
+
+
+## Downloading the model
+
+After installing Sample-Factory, download the model with:
+```
+python -m sample_factory.huggingface.load_from_hub -r qgallouedec/sample-factory-stick-pull-v2
+```
+
+    
+## Using the model
+
+To run the model after download, use the `enjoy` script corresponding to this environment:
+```
+python -m enjoy --algo=APPO --env=stick-pull-v2 --train_dir=./train_dir --experiment=sample-factory-stick-pull-v2
+```
+
+
+You can also upload models to the Hugging Face Hub using the same script with the `--push_to_hub` flag.
+See https://www.samplefactory.dev/10-huggingface/huggingface/ for more details
+    
+## Training with this model
+
+To continue training with this model, use the `train` script corresponding to this environment:
+```
+python -m train --algo=APPO --env=stick-pull-v2 --train_dir=./train_dir --experiment=sample-factory-stick-pull-v2 --restart_behavior=resume --train_for_env_steps=10000000000
+```
+
+Note, you may have to adjust `--train_for_env_steps` to a suitably high number as the experiment will resume at the number of steps it concluded at.
+    

checkpoint_p0/best_000035992_18427904_reward_96.670.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:97b7fe47e87bb203bc751ac344388c7559bfc5c86254bc8f1094d181da9aa3e8
+size 98239

checkpoint_p0/checkpoint_000195096_99889152.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d88ff88a83ac06fa42ef40e29713d48cdf663021e99fe025806ddcdea05992fb
+size 98567

checkpoint_p0/checkpoint_000195328_100007936.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e81e33041dedc191f09e8e404a725b0de4b019a81c94a3a1192061af1cc67fab
+size 98567

config.json

@@ -0,0 +1,134 @@
+{
+  "help": false,
+  "algo": "APPO",
+  "env": "stick-pull-v2",
+  "experiment": "stick-pull-v2",
+  "train_dir": "/home/qgallouedec/gia/data/envs/metaworld/train_dir",
+  "restart_behavior": "restart",
+  "device": "cpu",
+  "seed": null,
+  "num_policies": 1,
+  "async_rl": false,
+  "serial_mode": false,
+  "batched_sampling": false,
+  "num_batches_to_accumulate": 2,
+  "worker_num_splits": 2,
+  "policy_workers_per_policy": 1,
+  "max_policy_lag": 1000,
+  "num_workers": 8,
+  "num_envs_per_worker": 8,
+  "batch_size": 1024,
+  "num_batches_per_epoch": 4,
+  "num_epochs": 2,
+  "rollout": 64,
+  "recurrence": 1,
+  "shuffle_minibatches": false,
+  "gamma": 0.99,
+  "reward_scale": 0.1,
+  "reward_clip": 1000.0,
+  "value_bootstrap": true,
+  "normalize_returns": true,
+  "exploration_loss_coeff": 0.0,
+  "value_loss_coeff": 1.3,
+  "kl_loss_coeff": 0.1,
+  "exploration_loss": "entropy",
+  "gae_lambda": 0.95,
+  "ppo_clip_ratio": 0.2,
+  "ppo_clip_value": 1.0,
+  "with_vtrace": false,
+  "vtrace_rho": 1.0,
+  "vtrace_c": 1.0,
+  "optimizer": "adam",
+  "adam_eps": 1e-06,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.999,
+  "max_grad_norm": 3.5,
+  "learning_rate": 0.00295,
+  "lr_schedule": "linear_decay",
+  "lr_schedule_kl_threshold": 0.008,
+  "lr_adaptive_min": 1e-06,
+  "lr_adaptive_max": 0.01,
+  "obs_subtract_mean": 0.0,
+  "obs_scale": 1.0,
+  "normalize_input": true,
+  "normalize_input_keys": null,
+  "decorrelate_experience_max_seconds": 0,
+  "decorrelate_envs_on_one_worker": true,
+  "actor_worker_gpus": [],
+  "set_workers_cpu_affinity": true,
+  "force_envs_single_thread": false,
+  "default_niceness": 0,
+  "log_to_file": true,
+  "experiment_summaries_interval": 3,
+  "flush_summaries_interval": 30,
+  "stats_avg": 100,
+  "summaries_use_frameskip": true,
+  "heartbeat_interval": 20,
+  "heartbeat_reporting_interval": 180,
+  "train_for_env_steps": 100000000,
+  "train_for_seconds": 10000000000,
+  "save_every_sec": 15,
+  "keep_checkpoints": 2,
+  "load_checkpoint_kind": "latest",
+  "save_milestones_sec": -1,
+  "save_best_every_sec": 5,
+  "save_best_metric": "reward",
+  "save_best_after": 100000,
+  "benchmark": false,
+  "encoder_mlp_layers": [
+    64,
+    64
+  ],
+  "encoder_conv_architecture": "convnet_simple",
+  "encoder_conv_mlp_layers": [
+    512
+  ],
+  "use_rnn": false,
+  "rnn_size": 512,
+  "rnn_type": "gru",
+  "rnn_num_layers": 1,
+  "decoder_mlp_layers": [],
+  "nonlinearity": "tanh",
+  "policy_initialization": "torch_default",
+  "policy_init_gain": 1.0,
+  "actor_critic_share_weights": true,
+  "adaptive_stddev": false,
+  "continuous_tanh_scale": 0.0,
+  "initial_stddev": 1.0,
+  "use_env_info_cache": false,
+  "env_gpu_actions": false,
+  "env_gpu_observations": true,
+  "env_frameskip": 1,
+  "env_framestack": 1,
+  "pixel_format": "CHW",
+  "use_record_episode_statistics": false,
+  "with_wandb": true,
+  "wandb_user": "qgallouedec",
+  "wandb_project": "sample_facotry_metaworld",
+  "wandb_group": null,
+  "wandb_job_type": "SF",
+  "wandb_tags": [],
+  "with_pbt": false,
+  "pbt_mix_policies_in_one_env": true,
+  "pbt_period_env_steps": 5000000,
+  "pbt_start_mutation": 20000000,
+  "pbt_replace_fraction": 0.3,
+  "pbt_mutation_rate": 0.15,
+  "pbt_replace_reward_gap": 0.1,
+  "pbt_replace_reward_gap_absolute": 1e-06,
+  "pbt_optimize_gamma": false,
+  "pbt_target_objective": "true_objective",
+  "pbt_perturb_min": 1.1,
+  "pbt_perturb_max": 1.5,
+  "command_line": "--env stick-pull-v2 --experiment stick-pull-v2 --with_wandb True --wandb_user qgallouedec --wandb_project sample_facotry_metaworld",
+  "cli_args": {
+    "env": "stick-pull-v2",
+    "experiment": "stick-pull-v2",
+    "with_wandb": true,
+    "wandb_user": "qgallouedec",
+    "wandb_project": "sample_facotry_metaworld"
+  },
+  "git_hash": "2bb8d8c9cd813ffeafaad42038b1e71364092d36",
+  "git_repo_name": "https://github.com/huggingface/gia",
+  "wandb_unique_id": "stick-pull-v2_20230311_151346_189754"
+}

git.diff

@@ -0,0 +1,215 @@
+diff --git a/data/envs/metaworld/generate_dataset.py b/data/envs/metaworld/generate_dataset.py
+index dc51b4f..ed99a4a 100644
+--- a/data/envs/metaworld/generate_dataset.py
++++ b/data/envs/metaworld/generate_dataset.py
+@@ -4,7 +4,9 @@ from typing import Dict, Optional
+ import gym
+ import metaworld
+ import numpy as np
++import pandas as pd
+ import torch
++from datasets import Dataset
+ from huggingface_hub import HfApi, repocard, upload_folder
+ from sample_factory.algo.learning.learner import Learner
+ from sample_factory.algo.sampling.batched_sampling import preprocess_actions
+@@ -12,11 +14,7 @@ from sample_factory.algo.utils.action_distributions import argmax_actions
+ from sample_factory.algo.utils.env_info import extract_env_info
+ from sample_factory.algo.utils.make_env import make_env_func_batched
+ from sample_factory.algo.utils.rl_utils import make_dones, prepare_and_normalize_obs
+-from sample_factory.cfg.arguments import (
+-    load_from_checkpoint,
+-    parse_full_cfg,
+-    parse_sf_args,
+-)
++from sample_factory.cfg.arguments import load_from_checkpoint, parse_full_cfg, parse_sf_args
+ from sample_factory.envs.env_utils import register_env
+ from sample_factory.model.actor_critic import create_actor_critic
+ from sample_factory.model.model_utils import get_rnn_size
+@@ -165,10 +163,8 @@ def create_dataset(cfg: Config):
+     # Create dataset
+     dataset_size = 100_000
+     dataset = {
+-        "observations": np.zeros(
+-            (dataset_size, *env.observation_space["obs"].shape), dtype=env.observation_space["obs"].dtype
+-        ),
+-        "actions": np.zeros((dataset_size, *env.action_space.shape), env.action_space.dtype),
++        "observations": np.zeros((dataset_size, *env.observation_space["obs"].shape), dtype=np.float32),
++        "actions": np.zeros((dataset_size, *env.action_space.shape), np.float32),
+         "dones": np.zeros((dataset_size,), bool),
+         "rewards": np.zeros((dataset_size,), np.float32),
+     }
+@@ -206,6 +202,13 @@ def create_dataset(cfg: Config):
+ 
+     env.close()
+ 
++    # Convert dict of numpy array to pandas dataframe
++#    dataset = Dataset.from_dict(dataset)
++#    dataset.create_config_id
++    # Set the card of the dataset
++#    dataset.card = f""""""
++#    dataset.push_to_hub("qgallouedec/prj_gia_dataset_metaworld_assembly_v2_1111_demo")
++
+     # Save dataset
+     repo_path = f"{cfg.train_dir}/datasets/{cfg.experiment}"
+     os.makedirs(repo_path, exist_ok=True)
+diff --git a/data/envs/metaworld/generate_dataset_all.sh b/data/envs/metaworld/generate_dataset_all.sh
+index 802bf5c..3cc4f97 100755
+--- a/data/envs/metaworld/generate_dataset_all.sh
++++ b/data/envs/metaworld/generate_dataset_all.sh
+@@ -1,34 +1,6 @@
+ #!/bin/bash
+ 
+ ENVS=(
+-    assembly
+-    basketball
+-    bin-picking
+-    box-close
+-    button-press-topdown
+-    button-press-topdown-wall
+-    button-press
+-    button-press-wall
+-    coffee-button
+-    coffee-pull
+-    coffee-push
+-    dial-turn
+-    disassemble
+-    door-close
+-    door-lock
+-    door-open
+-    door-unlock
+-    drawer-close
+-    drawer-open
+-    faucet-close
+-    faucet-open
+-    hammer
+-    hand-insert
+-    handle-press-side
+-    handle-press
+-    handle-pull-side
+-    handle-pull
+-    lever-pull
+     peg-insert-side
+     peg-unplug-side
+     pick-out-of-hole
+@@ -40,19 +12,8 @@ ENVS=(
+     plate-slide
+     push-back
+     push
+-    push-wall
+-    reach
+-    reach-wall
+-    shelf-place
+-    soccer
+-    stick-pull
+-    stick-push
+-    sweep-into
+-    sweep
+-    window-close
+-    window-open
+ )
+ 
+ for ENV in "${ENVS[@]}"; do
+-    python generate_dataset.py --env $ENV-v2 --experiment $ENV-v2 --train_dir=./train_dir 
++    python generate_dataset.py --env $ENV-v2 --experiment $ENV-v2 --train_dir=./train_dir
+ done
+diff --git a/data/envs/metaworld/train_all.sh b/data/envs/metaworld/train_all.sh
+index dbf328a..1237f2c 100755
+--- a/data/envs/metaworld/train_all.sh
++++ b/data/envs/metaworld/train_all.sh
+@@ -1,48 +1,48 @@
+ #!/bin/bash
+ 
+ ENVS=(
+-    assembly
+-    basketball
+-    bin-picking
+-    box-close
+-    button-press-topdown
+-    button-press-topdown-wall
+-    button-press
+-    button-press-wall
+-    coffee-button
+-    coffee-pull
+-    coffee-push
+-    dial-turn
+-    disassemble
+-    door-close
+-    door-lock
+-    door-open
+-    door-unlock
+-    drawer-close
+-    drawer-open
+-    faucet-close
+-    faucet-open
+-    hammer
+-    hand-insert
+-    handle-press-side
+-    handle-press
+-    handle-pull-side
+-    handle-pull
+-    lever-pull
+-    peg-insert-side
+-    peg-unplug-side
+-    pick-out-of-hole
+-    pick-place
+-    pick-place-wall
+-    plate-slide-back-side
+-    plate-slide-back
+-    plate-slide-side
+-    plate-slide
+-    push-back
+-    push
+-    push-wall
+-    reach
+-    reach-wall
++#    assembly
++#   basketball
++#    bin-picking
++#    box-close
++#    button-press-topdown
++#    button-press-topdown-wall
++#    button-press
++#    button-press-wall
++#    coffee-button
++#    coffee-pull
++#    coffee-push
++#    dial-turn
++#    disassemble
++#    door-close
++#    door-lock
++#    door-open
++#    door-unlock
++#    drawer-close
++#    drawer-open
++#    faucet-close
++#    faucet-open
++#    hammer
++#    hand-insert
++#    handle-press-side
++#    handle-press
++#    handle-pull-side
++#    handle-pull
++#    lever-pull
++#    peg-insert-side
++#    peg-unplug-side
++#    pick-out-of-hole
++#    pick-place
++#    pick-place-wall
++#    plate-slide-back-side
++#    plate-slide-back
++#    plate-slide-side
++#    plate-slide
++#    push-back
++#    push
++#    push-wall
++#    reach
++#    reach-wall
+     shelf-place
+     soccer
+     stick-pull
+@@ -55,4 +55,5 @@ ENVS=(
+ 
+ for ENV in "${ENVS[@]}"; do
+     python train.py --env $ENV-v2 --experiment $ENV-v2 --with_wandb True --wandb_user qgallouedec --wandb_project sample_facotry_metaworld
++    python enjoy.py --algo=APPO --env $ENV-v2 --experiment $ENV-v2 --train_dir=./train_dir --max_num_episodes=10 --push_to_hub --hf_repository=qgallouedec/sample-factory-$ENV-v2 --save_video --no_render --enjoy_script=enjoy --train_script=train --load_checkpoint_kind best
+ done

replay.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24466ce525dd425a9ca92dabbcc653583fe24402022029bb9ea58ded3c56c9b8
+size 2694988