davidquarel commited on
Commit
3d11090
·
verified ·
1 Parent(s): 4a50986

Upload folder using huggingface_hub

Browse files
al_0.6_g_0.98_id_19_seed_980619/action_logprobs.pth.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:153913c47ce7e9ed7654333b9579b7a7b7ce3a139ed538b240024bd1fa6e5568
3
+ size 206068536
al_0.6_g_0.98_id_19_seed_980619/action_probs.tar.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:238bfe401420452660d71eb7f688c129a62cfe033f1c766a91a129e16d19c861
3
+ size 88940130
al_0.6_g_0.98_id_19_seed_980619/checkpoints.tar.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9bd534aa2af1a24f292ac0904ebc2b183631c8eb9bb526cb4e7b166986a36ae0
3
+ size 1538397302
al_0.6_g_0.98_id_19_seed_980619/config.cfg ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ rl_action=train
2
+ model_type=impala
3
+ num_rollout_steps=64
4
+ lr=5e-05
5
+ discount_rate=0.98
6
+ eff_horizon=None
7
+ eval_every=1
8
+ use_wandb=True
9
+ use_hf=True
10
+ use_log=True
11
+ log_optimizer_state=True
12
+ num_total_env_steps=10000000000
13
+ checkpoint=al_0.6_g_0.98_id_19_seed_980619
14
+ render_sixel=False
15
+ sixel_idx=60
16
+ seed=980619
17
+ run_id=19
18
+ seed_formula={int(discount_rate*100):02d}{int(alpha*10):02d}{run_id:02d}
19
+ mask_type=first_episode
20
+ penalize_time=False
21
+ optim=adam
22
+ live_monitor=False
23
+ use_bf16=False
24
+ deterministic=True
25
+ eval_schedule=0:1,250:2,500:5,2000:10
26
+ grad_acc_per_chunk=5
27
+ num_rollout_chunks=1
28
+ cheese_loc=any
29
+ env_rule=None
30
+ env_layout=open
31
+ alpha=0.6
32
+ env_size=13
33
+ num_levels=9600
34
+ f_str_ckpt=al_{alpha}_g_{discount_rate}_id_{run_id}_seed_{seed}
35
+ wandb_project=jaxgmg2_3phase_optim_state
36
+ ckpt_dir=jaxgmg2_3phase_optim_state
37
+ duplication_factor=-1
38
+ smoke=False
39
+ compile=True
40
+ num_chains=6
41
+ num_draws=3000
42
+ num_steps_bw_draws=1
43
+ on_policy=True
44
+ llc_nbeta=3000
45
+ localization=10
46
+ exact_solver_each_draw=False
47
+ llc_optimizer=sgld
48
+ iw_clip_eps=None
49
+ rmsprop_burnin_steps=20
50
+ llc_data_file=llc_scan_open_reinforce.pkl
51
+ llc_checkpoint_index=None
52
+ llc_checkpoint_number=None
53
+ sink=None
54
+ repo_id=davidquarel/jaxgmg_ckpt_zip
55
+ use_shuffled_checkpoints=False
56
+ force_re_download=False
57
+ off_distribution_data=False
58
+ weight_restrictions=None
59
+ weight_restrictions_invert=False
60
+ evaluate_every_position=False
61
+ num_prev_actions=1
62
+ ntfy=david_jaxgmg
63
+ vis_average_state=False
64
+ trim_episodes=False
65
+ use_prev_action=False
66
+ ckpt_path=jaxgmg2_3phase_optim_state/al_0.6_g_0.98_id_19_seed_980619
67
+ env_steps_per_loop=614400
68
+ total_loops=16276
69
+ eff_acc_steps=5
70
+ env_steps_per_microbatch=122880
71
+ chunk_size=9600
al_0.6_g_0.98_id_19_seed_980619/config.json ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha": 0.6,
3
+ "checkpoint": "al_0.6_g_0.98_id_19_seed_980619",
4
+ "cheese_loc": "any",
5
+ "chunk_size": 9600,
6
+ "ckpt_dir": "jaxgmg2_3phase_optim_state",
7
+ "ckpt_path": "jaxgmg2_3phase_optim_state/al_0.6_g_0.98_id_19_seed_980619",
8
+ "compile": true,
9
+ "deterministic": true,
10
+ "discount_rate": 0.98,
11
+ "duplication_factor": -1,
12
+ "eff_acc_steps": 5,
13
+ "eff_horizon": null,
14
+ "env_layout": "open",
15
+ "env_rule": null,
16
+ "env_size": 13,
17
+ "env_steps_per_loop": 614400,
18
+ "env_steps_per_microbatch": 122880,
19
+ "eval_every": 1,
20
+ "eval_schedule": "0:1,250:2,500:5,2000:10",
21
+ "evaluate_every_position": false,
22
+ "exact_solver_each_draw": false,
23
+ "f_str_ckpt": "al_{alpha}_g_{discount_rate}_id_{run_id}_seed_{seed}",
24
+ "force_re_download": false,
25
+ "grad_acc_per_chunk": 5,
26
+ "iw_clip_eps": null,
27
+ "live_monitor": false,
28
+ "llc_checkpoint_index": null,
29
+ "llc_checkpoint_number": null,
30
+ "llc_data_file": "llc_scan_open_reinforce.pkl",
31
+ "llc_nbeta": 3000,
32
+ "llc_optimizer": "sgld",
33
+ "localization": 10,
34
+ "log_optimizer_state": true,
35
+ "lr": 5e-05,
36
+ "mask_type": "first_episode",
37
+ "model_type": "impala",
38
+ "ntfy": "david_jaxgmg",
39
+ "num_chains": 6,
40
+ "num_draws": 3000,
41
+ "num_levels": 9600,
42
+ "num_prev_actions": 1,
43
+ "num_rollout_chunks": 1,
44
+ "num_rollout_steps": 64,
45
+ "num_steps_bw_draws": 1,
46
+ "num_total_env_steps": 10000000000,
47
+ "off_distribution_data": false,
48
+ "on_policy": true,
49
+ "optim": "adam",
50
+ "penalize_time": false,
51
+ "render_sixel": false,
52
+ "repo_id": "davidquarel/jaxgmg_ckpt_zip",
53
+ "rl_action": "train",
54
+ "rmsprop_burnin_steps": 20,
55
+ "run_id": 19,
56
+ "seed": 980619,
57
+ "seed_formula": "{int(discount_rate*100):02d}{int(alpha*10):02d}{run_id:02d}",
58
+ "sink": null,
59
+ "sixel_idx": 60,
60
+ "smoke": false,
61
+ "total_loops": 16276,
62
+ "trim_episodes": false,
63
+ "use_bf16": false,
64
+ "use_hf": true,
65
+ "use_log": true,
66
+ "use_prev_action": false,
67
+ "use_shuffled_checkpoints": false,
68
+ "use_wandb": true,
69
+ "vis_average_state": false,
70
+ "wandb_project": "jaxgmg2_3phase_optim_state",
71
+ "weight_restrictions": null,
72
+ "weight_restrictions_invert": false
73
+ }
al_0.6_g_0.98_id_19_seed_980619/eval.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
al_0.6_g_0.98_id_19_seed_980619/eval.log ADDED
The diff for this file is too large to render. See raw diff
 
al_0.6_g_0.98_id_19_seed_980619/latest_train.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "ts": "2026-02-13T04:11:06.888898Z",
3
+ "loop": 16275,
4
+ "env_steps": 9999974963,
5
+ "loss": 0.02195007912814617,
6
+ "avg_return": 0.8617340326309204,
7
+ "regret": 0.0009366097510792315,
8
+ "entropy": 0.10705839097499847,
9
+ "lr": 5e-05,
10
+ "discount_rate": 0.98,
11
+ "num_rollout_steps": 64
12
+ }
al_0.6_g_0.98_id_19_seed_980619/optimizer_state.tar.gz ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59149c3571ca03fb127e4af832ced2671d59ceb40904631d30d90c76a3935261
3
+ size 2549312061
al_0.6_g_0.98_id_19_seed_980619/setup.json ADDED
@@ -0,0 +1,88 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "rl_action": "train",
4
+ "model_type": "impala",
5
+ "num_rollout_steps": 64,
6
+ "lr": 5e-05,
7
+ "discount_rate": 0.98,
8
+ "eff_horizon": null,
9
+ "eval_every": 1,
10
+ "use_wandb": true,
11
+ "use_hf": true,
12
+ "use_log": true,
13
+ "log_optimizer_state": true,
14
+ "num_total_env_steps": 10000000000,
15
+ "checkpoint": "al_0.6_g_0.98_id_19_seed_980619",
16
+ "render_sixel": false,
17
+ "sixel_idx": 60,
18
+ "seed": 980619,
19
+ "run_id": 19,
20
+ "seed_formula": "{int(discount_rate*100):02d}{int(alpha*10):02d}{run_id:02d}",
21
+ "mask_type": "first_episode",
22
+ "penalize_time": false,
23
+ "optim": "adam",
24
+ "live_monitor": false,
25
+ "use_bf16": false,
26
+ "deterministic": true,
27
+ "eval_schedule": "0:1,250:2,500:5,2000:10",
28
+ "grad_acc_per_chunk": 5,
29
+ "num_rollout_chunks": 1,
30
+ "cheese_loc": "any",
31
+ "env_rule": null,
32
+ "env_layout": "open",
33
+ "alpha": 0.6,
34
+ "env_size": 13,
35
+ "num_levels": 9600,
36
+ "f_str_ckpt": "al_{alpha}_g_{discount_rate}_id_{run_id}_seed_{seed}",
37
+ "wandb_project": "jaxgmg2_3phase_optim_state",
38
+ "ckpt_dir": "jaxgmg2_3phase_optim_state",
39
+ "duplication_factor": -1,
40
+ "smoke": false,
41
+ "compile": true,
42
+ "num_chains": 6,
43
+ "num_draws": 3000,
44
+ "num_steps_bw_draws": 1,
45
+ "on_policy": true,
46
+ "llc_nbeta": 3000,
47
+ "localization": 10,
48
+ "exact_solver_each_draw": false,
49
+ "llc_optimizer": "sgld",
50
+ "iw_clip_eps": null,
51
+ "rmsprop_burnin_steps": 20,
52
+ "llc_data_file": "llc_scan_open_reinforce.pkl",
53
+ "llc_checkpoint_index": null,
54
+ "llc_checkpoint_number": null,
55
+ "sink": null,
56
+ "repo_id": "davidquarel/jaxgmg_ckpt_zip",
57
+ "use_shuffled_checkpoints": false,
58
+ "force_re_download": false,
59
+ "off_distribution_data": false,
60
+ "weight_restrictions": null,
61
+ "weight_restrictions_invert": false,
62
+ "evaluate_every_position": false,
63
+ "num_prev_actions": 1,
64
+ "ntfy": "david_jaxgmg",
65
+ "vis_average_state": false,
66
+ "trim_episodes": false,
67
+ "use_prev_action": false,
68
+ "ckpt_path": "jaxgmg2_3phase_optim_state/al_0.6_g_0.98_id_19_seed_980619",
69
+ "env_steps_per_loop": 614400,
70
+ "total_loops": 16276,
71
+ "eff_acc_steps": 5,
72
+ "env_steps_per_microbatch": 122880,
73
+ "chunk_size": 9600
74
+ },
75
+ "setup": {
76
+ "start_time_utc": "2026-02-13T00:08:39.238936Z",
77
+ "seed": 980619,
78
+ "device": "cuda",
79
+ "python_version": "3.11.11",
80
+ "torch_version": "2.10.0+cu128",
81
+ "jax_version": "0.6.2",
82
+ "hostname": "65e7cc5b871a",
83
+ "platform": "Linux-6.5.0-35-generic-x86_64-with-glibc2.35",
84
+ "git_sha": null,
85
+ "git_branch": null,
86
+ "use_wandb": true
87
+ }
88
+ }
al_0.6_g_0.98_id_19_seed_980619/setup.txt ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ Start: 2026-02-13T00:08:39.238936Z
2
+ Seed: 980619
3
+ Device: cuda
4
+ Python: 3.11.11 | Torch: 2.10.0+cu128 | JAX: 0.6.2
5
+ Host: 65e7cc5b871a
6
+ Platform: Linux-6.5.0-35-generic-x86_64-with-glibc2.35
7
+ Git: branch=None sha=None
8
+ use_wandb: True
al_0.6_g_0.98_id_19_seed_980619/train.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
al_0.6_g_0.98_id_19_seed_980619/train.log ADDED
The diff for this file is too large to render. See raw diff