small_scale_iql_chaotic_lstm_multiseed-v0 | tlab Workspace

Howuhh's group workspace

Group: small_scale_iql_chaotic_lstm_multiseed-v0

114

1-1

of 1

Tags

Notes

Author

howuhh

State

Finished

Start time

June 6th, 2023 5:06:15 PM

Sweep

ejhec3rp

Runtime

14h 49m 26s

Tracked hours

14h 48m 48s

Run path

tlab/NetHack/wt4mfbin

Linux-5.4.0-147-generic-x86_64-with-glibc2.29

Python version

3.8.10

Git repository

git clone git@github.com:tinkoff-ai/katakomba.git

Git state

git checkout -b "iql-ran-elf-cha-03d93382" 64b391eace6509b40575bb48631df003afd695d0

Command

algorithms/small_scale/iql_chaotic_lstm.py --character=ran-elf-cha --data_mode=memmap --group=small_scale_iql_chaotic_lstm_multiseed --train_seed=0 --version=0

System Hardware

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

W&B CLI Version

0.12.21

Group

small_scale_iql_chaotic_lstm_multiseed-v0

Config parameters are your model's inputs. Learn more

▶
Config parameters:{} 27 keys
- batch_size:
  64
- character:
  "ran-elf-cha"
- checkpoints_path:
  null
- clip_grad_norm:
  null
- clip_range:
  10
- data_mode:
  "memmap"
- eval_episodes:
  50
- eval_every:
  10,000
- eval_processes:
  14
- eval_seed:
  50
- expectile_tau:
  0.8
- gamma:
  0.999
- group:
  "small_scale_iql_chaotic_lstm_multiseed"
- learning_rate:
  0.0003
- name:
  "iql-ran-elf-cha-03d93382"
- project:
  "NetHack"
- rnn_dropout:
  0
- rnn_hidden_dim:
  2,048
- rnn_layers:
  2
- seq_len:
  16
- tau:
  0.005
- temperature:
  1
- train_seed:
  0
- update_steps:
  500,000
- use_prev_action:
  true
- version:
  0
- weight_decay:
  0

Summary metrics are your model's outputs. Learn more

▶
Summary metrics:{} 23 keys
- actor_loss:
  0.23648372292518616
- depth_max:
  1
- depth_mean:
  0.02
- depth_median:
  0
- depth_min:
  0
- depth_std:
  0.14
- loss:
  0.2388196885585785
- next_v:
  14.921875
- q_target:
  14.911907196044922
- reward_max:
  3,584
- reward_mean:
  618.38
- reward_median:
  475
- reward_min:
  2
- reward_std:
  679.1983183135835
- td_loss:
  0.002041602274402976
- times/backward_pass:
  0.014640735626220704
- times/batch_loading_cpu:
  0.0681757926940918
- times/batch_loading_gpu:
  0.06811673736572266
- times/evaluation_cpu:
  101.29734778404236
- times/evaluation_gpu:
  101.296703125
- times/forward_pass:
  0.013387999534606933
- transitions:
  512,000,000
- value_loss:
  0.0002943634754046798

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

wandb-history

run-wt4mfbin-history:v0