state-machine

State-machine's group workspace

Group: XLand-MiniGrid-R1-13x13-img-sweep-v2

1-20

of 20

Tags

Notes

Author

state-machine

State

Finished

Start time

March 14th, 2024 5:16:22 AM

Sweep

frq53gtm

Runtime

8h 46m 23s

Tracked hours

8h 46m 11s

Run path

state-machine/xminigrid/18dz6tls

Linux-5.4.0-171-generic-x86_64-with-glibc2.31

Python version

3.10.11

Git repository

git clone git@github.com:corl-team/xland-minigrid.git

Git state

git checkout -b "meta-task-ppo" dc26721ec7b30dd9c4a4a7c6ee3fb3c7fcf69d32

Command

/workspace/xminigrid/training/train_meta_task.py --benchmark_id=high-1m --config_path=configs/config_best_img.yaml --group=XLand-MiniGrid-R1-13x13-img-sweep-v2 --train_seed=5

System Hardware

CPU count	64
Logical CPU count	128
GPU count	1
GPU type	NVIDIA A100-PCIE-40GB

W&B CLI Version

0.15.10

Group

XLand-MiniGrid-R1-13x13-img-sweep-v2

Config parameters are your model's inputs. Learn more

▶
Config parameters:{} 29 keys
- action_emb_dim:
  16
- benchmark_id:
  "high-1m"
- checkpoint_path:
  null
- clip_eps:
  0.2
- config_path:
  "configs/config_best_img.yaml"
- ent_coef:
  0.01
- env_id:
  "XLand-MiniGrid-R1-13x13"
- eval_num_envs:
  8,192
- eval_num_episodes:
  25
- eval_seed:
  42
- gae_lambda:
  0.95
- gamma:
  0.99
- group:
  "XLand-MiniGrid-R1-13x13-img-sweep-v2"
- head_hidden_dim:
  256
- img_obs:
  true
- lr:
  0.0001
- max_grad_norm:
  0.5
- name:
  "meta-task-ppo"
- num_envs:
  1,024
- num_minibatches:
  32
- num_steps_per_env:
  12,800
- num_steps_per_update:
  128
- project:
  "xminigrid"
- rnn_hidden_dim:
  1,024
- rnn_num_layers:
  1
- total_timesteps:
  500,000,000
- train_seed:
  5
- update_epochs:
  1
- vf_coef:
  0.5

Summary metrics are your model's outputs. Learn more

▶
Summary metrics:{} 13 keys
- actor_loss:
  -0.00000490473121317336
- entropy:
  1.6150246858596802
- eval/lengths:
  12,114.16796875
- eval/lengths_20percentile:
  12,566.2001953125
- eval/returns_20percentile:
  0
- eval/returns_mean:
  1.1862820386886597
- eval/returns_median:
  0
- lr:
  0.00000263158085545001
- steps_per_second:
  15,873.255760692648
- total_loss:
  -0.015854911878705025
- training_time:
  31,499.523950099945
- transitions:
  498,073,600
- value_loss:
  0.000600480183493346

This run produced these artifacts as outputs. Total: 1. Learn more

wandb-history

run-18dz6tls-history:v0