AbhishekGhosh1998 commited on
Commit
d719d62
·
verified ·
1 Parent(s): 5781f44

granularity results ultraep/tierbmem @ 2026-08-26T19:31:45Z from sp-aiinfra-interactive-aghosh-b5xzj-8xb200-moe-cg-0 (part 4)

Browse files
Files changed (13) hide show
  1. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0006_iter000004_20260826T192929276544Z.pickle +3 -0
  2. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000000_20260826T192927662362Z.json +79 -0
  3. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000000_20260826T192927662362Z.pickle +3 -0
  4. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000001_20260826T192928418621Z.json +79 -0
  5. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000001_20260826T192928418621Z.pickle +3 -0
  6. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000002_20260826T192928707580Z.json +79 -0
  7. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000002_20260826T192928707580Z.pickle +3 -0
  8. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000003_20260826T192928992983Z.json +79 -0
  9. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000003_20260826T192928992983Z.pickle +3 -0
  10. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000004_20260826T192929276583Z.json +79 -0
  11. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000004_20260826T192929276583Z.pickle +3 -0
  12. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/run.env +61 -0
  13. ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/train.log +1221 -0
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0006_iter000004_20260826T192929276544Z.pickle ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:259954976480f1a5bf25fbbbabe10fd34585802adab4575007be73ce476c3f97
3
+ size 1183504
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000000_20260826T192927662362Z.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "active_bytes_current": 67504251904,
3
+ "active_bytes_current_delta": 26818203648,
4
+ "active_bytes_peak": 68365347840,
5
+ "active_bytes_peak_delta": 27641550848,
6
+ "after": {
7
+ "active_bytes_current": 67504251904,
8
+ "active_bytes_peak": 68365347840,
9
+ "allocated_bytes_current": 67504251904,
10
+ "allocated_bytes_peak": 68365347840,
11
+ "device": 7,
12
+ "inactive_split_bytes_current": 2089737216,
13
+ "inactive_split_bytes_peak": 4508824064,
14
+ "iteration": 0,
15
+ "mode": "iteration",
16
+ "num_alloc_retries": 0,
17
+ "num_ooms": 0,
18
+ "phase": "iteration_end",
19
+ "rank": 7,
20
+ "reserved_bytes_current": 77219233792,
21
+ "reserved_bytes_peak": 77219233792,
22
+ "segment_count_current": 980,
23
+ "segment_count_peak": 980,
24
+ "timestamp_utc": "2026-08-26T19:29:27.661854Z",
25
+ "world_size": 8
26
+ },
27
+ "allocated_bytes_current": 67504251904,
28
+ "allocated_bytes_current_delta": 26818203648,
29
+ "allocated_bytes_peak": 68365347840,
30
+ "allocated_bytes_peak_delta": 27641550848,
31
+ "before": {
32
+ "active_bytes_current": 40686048256,
33
+ "active_bytes_peak": 40723796992,
34
+ "allocated_bytes_current": 40686048256,
35
+ "allocated_bytes_peak": 40723796992,
36
+ "device": 7,
37
+ "inactive_split_bytes_current": 508308480,
38
+ "inactive_split_bytes_peak": 728509440,
39
+ "iteration": 0,
40
+ "mode": "iteration",
41
+ "num_alloc_retries": 0,
42
+ "num_ooms": 0,
43
+ "phase": "iteration_start",
44
+ "rank": 7,
45
+ "reserved_bytes_current": 43864031232,
46
+ "reserved_bytes_peak": 43864031232,
47
+ "segment_count_current": 261,
48
+ "segment_count_peak": 261,
49
+ "timestamp_utc": "2026-08-26T19:28:59.093442Z",
50
+ "world_size": 8
51
+ },
52
+ "device": 7,
53
+ "device_delta": 0,
54
+ "inactive_split_bytes_current": 2089737216,
55
+ "inactive_split_bytes_current_delta": 1581428736,
56
+ "inactive_split_bytes_peak": 4508824064,
57
+ "inactive_split_bytes_peak_delta": 3780314624,
58
+ "iteration": 0,
59
+ "iteration_delta": 0,
60
+ "mode": "iteration",
61
+ "num_alloc_retries": 0,
62
+ "num_alloc_retries_delta": 0,
63
+ "num_ooms": 0,
64
+ "num_ooms_delta": 0,
65
+ "rank": 7,
66
+ "rank_delta": 0,
67
+ "reserved_bytes_current": 77219233792,
68
+ "reserved_bytes_current_delta": 33355202560,
69
+ "reserved_bytes_peak": 77219233792,
70
+ "reserved_bytes_peak_delta": 33355202560,
71
+ "segment_count_current": 980,
72
+ "segment_count_current_delta": 719,
73
+ "segment_count_peak": 980,
74
+ "segment_count_peak_delta": 719,
75
+ "snapshot_pickle": "snapshot_iteration_rank0007_iter000000_20260826T192927662362Z.pickle",
76
+ "timestamp_utc": "2026-08-26T19:29:27.689794Z",
77
+ "world_size": 8,
78
+ "world_size_delta": 0
79
+ }
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000000_20260826T192927662362Z.pickle ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d302ad7252d338b8b3572ef4c89df5affc5c8c8ae9dee75d245f4314427a75d
3
+ size 1448545
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000001_20260826T192928418621Z.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "active_bytes_current": 67504251904,
3
+ "active_bytes_current_delta": 0,
4
+ "active_bytes_peak": 89859424256,
5
+ "active_bytes_peak_delta": 21494076416,
6
+ "after": {
7
+ "active_bytes_current": 67504251904,
8
+ "active_bytes_peak": 89859424256,
9
+ "allocated_bytes_current": 67504251904,
10
+ "allocated_bytes_peak": 89859424256,
11
+ "device": 7,
12
+ "inactive_split_bytes_current": 2089737216,
13
+ "inactive_split_bytes_peak": 4508824064,
14
+ "iteration": 1,
15
+ "mode": "iteration",
16
+ "num_alloc_retries": 0,
17
+ "num_ooms": 0,
18
+ "phase": "iteration_end",
19
+ "rank": 7,
20
+ "reserved_bytes_current": 91460993024,
21
+ "reserved_bytes_peak": 91460993024,
22
+ "segment_count_current": 1080,
23
+ "segment_count_peak": 1080,
24
+ "timestamp_utc": "2026-08-26T19:29:28.418170Z",
25
+ "world_size": 8
26
+ },
27
+ "allocated_bytes_current": 67504251904,
28
+ "allocated_bytes_current_delta": 0,
29
+ "allocated_bytes_peak": 89859424256,
30
+ "allocated_bytes_peak_delta": 21494076416,
31
+ "before": {
32
+ "active_bytes_current": 67504251904,
33
+ "active_bytes_peak": 68365347840,
34
+ "allocated_bytes_current": 67504251904,
35
+ "allocated_bytes_peak": 68365347840,
36
+ "device": 7,
37
+ "inactive_split_bytes_current": 2089737216,
38
+ "inactive_split_bytes_peak": 4508824064,
39
+ "iteration": 1,
40
+ "mode": "iteration",
41
+ "num_alloc_retries": 0,
42
+ "num_ooms": 0,
43
+ "phase": "iteration_start",
44
+ "rank": 7,
45
+ "reserved_bytes_current": 77219233792,
46
+ "reserved_bytes_peak": 77219233792,
47
+ "segment_count_current": 980,
48
+ "segment_count_peak": 980,
49
+ "timestamp_utc": "2026-08-26T19:29:27.692736Z",
50
+ "world_size": 8
51
+ },
52
+ "device": 7,
53
+ "device_delta": 0,
54
+ "inactive_split_bytes_current": 2089737216,
55
+ "inactive_split_bytes_current_delta": 0,
56
+ "inactive_split_bytes_peak": 4508824064,
57
+ "inactive_split_bytes_peak_delta": 0,
58
+ "iteration": 1,
59
+ "iteration_delta": 0,
60
+ "mode": "iteration",
61
+ "num_alloc_retries": 0,
62
+ "num_alloc_retries_delta": 0,
63
+ "num_ooms": 0,
64
+ "num_ooms_delta": 0,
65
+ "rank": 7,
66
+ "rank_delta": 0,
67
+ "reserved_bytes_current": 91460993024,
68
+ "reserved_bytes_current_delta": 14241759232,
69
+ "reserved_bytes_peak": 91460993024,
70
+ "reserved_bytes_peak_delta": 14241759232,
71
+ "segment_count_current": 1080,
72
+ "segment_count_current_delta": 100,
73
+ "segment_count_peak": 1080,
74
+ "segment_count_peak_delta": 100,
75
+ "snapshot_pickle": "snapshot_iteration_rank0007_iter000001_20260826T192928418621Z.pickle",
76
+ "timestamp_utc": "2026-08-26T19:29:28.442605Z",
77
+ "world_size": 8,
78
+ "world_size_delta": 0
79
+ }
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000001_20260826T192928418621Z.pickle ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb0770731178b4e8fd708fb623a8c132bf59c2a0e2ff77b4b65b52d5b57c9dd4
3
+ size 1193149
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000002_20260826T192928707580Z.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "active_bytes_current": 67504251904,
3
+ "active_bytes_current_delta": 0,
4
+ "active_bytes_peak": 89859424256,
5
+ "active_bytes_peak_delta": 0,
6
+ "after": {
7
+ "active_bytes_current": 67504251904,
8
+ "active_bytes_peak": 89859424256,
9
+ "allocated_bytes_current": 67504251904,
10
+ "allocated_bytes_peak": 89859424256,
11
+ "device": 7,
12
+ "inactive_split_bytes_current": 2089737216,
13
+ "inactive_split_bytes_peak": 5742727680,
14
+ "iteration": 2,
15
+ "mode": "iteration",
16
+ "num_alloc_retries": 0,
17
+ "num_ooms": 0,
18
+ "phase": "iteration_end",
19
+ "rank": 7,
20
+ "reserved_bytes_current": 92285173760,
21
+ "reserved_bytes_peak": 92285173760,
22
+ "segment_count_current": 1081,
23
+ "segment_count_peak": 1081,
24
+ "timestamp_utc": "2026-08-26T19:29:28.707147Z",
25
+ "world_size": 8
26
+ },
27
+ "allocated_bytes_current": 67504251904,
28
+ "allocated_bytes_current_delta": 0,
29
+ "allocated_bytes_peak": 89859424256,
30
+ "allocated_bytes_peak_delta": 0,
31
+ "before": {
32
+ "active_bytes_current": 67504251904,
33
+ "active_bytes_peak": 89859424256,
34
+ "allocated_bytes_current": 67504251904,
35
+ "allocated_bytes_peak": 89859424256,
36
+ "device": 7,
37
+ "inactive_split_bytes_current": 2089737216,
38
+ "inactive_split_bytes_peak": 4508824064,
39
+ "iteration": 2,
40
+ "mode": "iteration",
41
+ "num_alloc_retries": 0,
42
+ "num_ooms": 0,
43
+ "phase": "iteration_start",
44
+ "rank": 7,
45
+ "reserved_bytes_current": 91460993024,
46
+ "reserved_bytes_peak": 91460993024,
47
+ "segment_count_current": 1080,
48
+ "segment_count_peak": 1080,
49
+ "timestamp_utc": "2026-08-26T19:29:28.445584Z",
50
+ "world_size": 8
51
+ },
52
+ "device": 7,
53
+ "device_delta": 0,
54
+ "inactive_split_bytes_current": 2089737216,
55
+ "inactive_split_bytes_current_delta": 0,
56
+ "inactive_split_bytes_peak": 5742727680,
57
+ "inactive_split_bytes_peak_delta": 1233903616,
58
+ "iteration": 2,
59
+ "iteration_delta": 0,
60
+ "mode": "iteration",
61
+ "num_alloc_retries": 0,
62
+ "num_alloc_retries_delta": 0,
63
+ "num_ooms": 0,
64
+ "num_ooms_delta": 0,
65
+ "rank": 7,
66
+ "rank_delta": 0,
67
+ "reserved_bytes_current": 92285173760,
68
+ "reserved_bytes_current_delta": 824180736,
69
+ "reserved_bytes_peak": 92285173760,
70
+ "reserved_bytes_peak_delta": 824180736,
71
+ "segment_count_current": 1081,
72
+ "segment_count_current_delta": 1,
73
+ "segment_count_peak": 1081,
74
+ "segment_count_peak_delta": 1,
75
+ "snapshot_pickle": "snapshot_iteration_rank0007_iter000002_20260826T192928707580Z.pickle",
76
+ "timestamp_utc": "2026-08-26T19:29:28.729037Z",
77
+ "world_size": 8,
78
+ "world_size_delta": 0
79
+ }
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000002_20260826T192928707580Z.pickle ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eaf244fb4c06fb66ee1111ca5902f303694f4d98ed204ebbc0b5d4eab7edc338
3
+ size 1185223
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000003_20260826T192928992983Z.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "active_bytes_current": 67504251904,
3
+ "active_bytes_current_delta": 0,
4
+ "active_bytes_peak": 89897571840,
5
+ "active_bytes_peak_delta": 38147584,
6
+ "after": {
7
+ "active_bytes_current": 67504251904,
8
+ "active_bytes_peak": 89897571840,
9
+ "allocated_bytes_current": 67504251904,
10
+ "allocated_bytes_peak": 89897571840,
11
+ "device": 7,
12
+ "inactive_split_bytes_current": 2089737216,
13
+ "inactive_split_bytes_peak": 5912684032,
14
+ "iteration": 3,
15
+ "mode": "iteration",
16
+ "num_alloc_retries": 0,
17
+ "num_ooms": 0,
18
+ "phase": "iteration_end",
19
+ "rank": 7,
20
+ "reserved_bytes_current": 93109354496,
21
+ "reserved_bytes_peak": 93109354496,
22
+ "segment_count_current": 1082,
23
+ "segment_count_peak": 1082,
24
+ "timestamp_utc": "2026-08-26T19:29:28.992554Z",
25
+ "world_size": 8
26
+ },
27
+ "allocated_bytes_current": 67504251904,
28
+ "allocated_bytes_current_delta": 0,
29
+ "allocated_bytes_peak": 89897571840,
30
+ "allocated_bytes_peak_delta": 38147584,
31
+ "before": {
32
+ "active_bytes_current": 67504251904,
33
+ "active_bytes_peak": 89859424256,
34
+ "allocated_bytes_current": 67504251904,
35
+ "allocated_bytes_peak": 89859424256,
36
+ "device": 7,
37
+ "inactive_split_bytes_current": 2089737216,
38
+ "inactive_split_bytes_peak": 5742727680,
39
+ "iteration": 3,
40
+ "mode": "iteration",
41
+ "num_alloc_retries": 0,
42
+ "num_ooms": 0,
43
+ "phase": "iteration_start",
44
+ "rank": 7,
45
+ "reserved_bytes_current": 92285173760,
46
+ "reserved_bytes_peak": 92285173760,
47
+ "segment_count_current": 1081,
48
+ "segment_count_peak": 1081,
49
+ "timestamp_utc": "2026-08-26T19:29:28.731561Z",
50
+ "world_size": 8
51
+ },
52
+ "device": 7,
53
+ "device_delta": 0,
54
+ "inactive_split_bytes_current": 2089737216,
55
+ "inactive_split_bytes_current_delta": 0,
56
+ "inactive_split_bytes_peak": 5912684032,
57
+ "inactive_split_bytes_peak_delta": 169956352,
58
+ "iteration": 3,
59
+ "iteration_delta": 0,
60
+ "mode": "iteration",
61
+ "num_alloc_retries": 0,
62
+ "num_alloc_retries_delta": 0,
63
+ "num_ooms": 0,
64
+ "num_ooms_delta": 0,
65
+ "rank": 7,
66
+ "rank_delta": 0,
67
+ "reserved_bytes_current": 93109354496,
68
+ "reserved_bytes_current_delta": 824180736,
69
+ "reserved_bytes_peak": 93109354496,
70
+ "reserved_bytes_peak_delta": 824180736,
71
+ "segment_count_current": 1082,
72
+ "segment_count_current_delta": 1,
73
+ "segment_count_peak": 1082,
74
+ "segment_count_peak_delta": 1,
75
+ "snapshot_pickle": "snapshot_iteration_rank0007_iter000003_20260826T192928992983Z.pickle",
76
+ "timestamp_utc": "2026-08-26T19:29:29.012947Z",
77
+ "world_size": 8,
78
+ "world_size_delta": 0
79
+ }
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000003_20260826T192928992983Z.pickle ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4833679b1ff88e2d50737a74681595ea4673628734a5c969f2cd14a5228ea368
3
+ size 1185198
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000004_20260826T192929276583Z.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "active_bytes_current": 67504251904,
3
+ "active_bytes_current_delta": 0,
4
+ "active_bytes_peak": 89909421568,
5
+ "active_bytes_peak_delta": 11849728,
6
+ "after": {
7
+ "active_bytes_current": 67504251904,
8
+ "active_bytes_peak": 89909421568,
9
+ "allocated_bytes_current": 67504251904,
10
+ "allocated_bytes_peak": 89909421568,
11
+ "device": 7,
12
+ "inactive_split_bytes_current": 2089737216,
13
+ "inactive_split_bytes_peak": 5912684032,
14
+ "iteration": 4,
15
+ "mode": "iteration",
16
+ "num_alloc_retries": 0,
17
+ "num_ooms": 0,
18
+ "phase": "iteration_end",
19
+ "rank": 7,
20
+ "reserved_bytes_current": 93109354496,
21
+ "reserved_bytes_peak": 93109354496,
22
+ "segment_count_current": 1082,
23
+ "segment_count_peak": 1082,
24
+ "timestamp_utc": "2026-08-26T19:29:29.276153Z",
25
+ "world_size": 8
26
+ },
27
+ "allocated_bytes_current": 67504251904,
28
+ "allocated_bytes_current_delta": 0,
29
+ "allocated_bytes_peak": 89909421568,
30
+ "allocated_bytes_peak_delta": 11849728,
31
+ "before": {
32
+ "active_bytes_current": 67504251904,
33
+ "active_bytes_peak": 89897571840,
34
+ "allocated_bytes_current": 67504251904,
35
+ "allocated_bytes_peak": 89897571840,
36
+ "device": 7,
37
+ "inactive_split_bytes_current": 2089737216,
38
+ "inactive_split_bytes_peak": 5912684032,
39
+ "iteration": 4,
40
+ "mode": "iteration",
41
+ "num_alloc_retries": 0,
42
+ "num_ooms": 0,
43
+ "phase": "iteration_start",
44
+ "rank": 7,
45
+ "reserved_bytes_current": 93109354496,
46
+ "reserved_bytes_peak": 93109354496,
47
+ "segment_count_current": 1082,
48
+ "segment_count_peak": 1082,
49
+ "timestamp_utc": "2026-08-26T19:29:29.015437Z",
50
+ "world_size": 8
51
+ },
52
+ "device": 7,
53
+ "device_delta": 0,
54
+ "inactive_split_bytes_current": 2089737216,
55
+ "inactive_split_bytes_current_delta": 0,
56
+ "inactive_split_bytes_peak": 5912684032,
57
+ "inactive_split_bytes_peak_delta": 0,
58
+ "iteration": 4,
59
+ "iteration_delta": 0,
60
+ "mode": "iteration",
61
+ "num_alloc_retries": 0,
62
+ "num_alloc_retries_delta": 0,
63
+ "num_ooms": 0,
64
+ "num_ooms_delta": 0,
65
+ "rank": 7,
66
+ "rank_delta": 0,
67
+ "reserved_bytes_current": 93109354496,
68
+ "reserved_bytes_current_delta": 0,
69
+ "reserved_bytes_peak": 93109354496,
70
+ "reserved_bytes_peak_delta": 0,
71
+ "segment_count_current": 1082,
72
+ "segment_count_current_delta": 0,
73
+ "segment_count_peak": 1082,
74
+ "segment_count_peak_delta": 0,
75
+ "snapshot_pickle": "snapshot_iteration_rank0007_iter000004_20260826T192929276583Z.pickle",
76
+ "timestamp_utc": "2026-08-26T19:29:29.296802Z",
77
+ "world_size": 8,
78
+ "world_size_delta": 0
79
+ }
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot_iteration_rank0007_iter000004_20260826T192929276583Z.pickle ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6213150d37196bec7f033fffc339599a9b3a5129cea1d460e310e5b7b2f5ea8
3
+ size 1185128
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/run.env ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ backend=ultraep
2
+ config=QWEN2_57B
3
+ family=tierB
4
+ repo=/workspace/Megatron-UltraEP
5
+ branch=ultraep-better-memory-dump-support
6
+ commit=c3a8a8a7
7
+ num_layers=10
8
+ hidden=3584
9
+ ffn_hidden=2560
10
+ moe_ffn_hidden=2560
11
+ num_experts=64
12
+ topk=8
13
+ experts_per_rank=8
14
+ expert_params=27525120
15
+ heads=28
16
+ kv_groups=4
17
+ kv_channels=
18
+ qk_layernorm=0
19
+ shared_ffn=20480
20
+ shared_expert_overlap=0
21
+ mla=0
22
+ dense_first_layers=0
23
+ moe_layers=10
24
+ rotary_base=1000000
25
+ rotary_scaling=
26
+ qkv_bias=1
27
+ router_balancing=aux_loss
28
+ aux_loss_coeff=1e-2
29
+ router_pre_softmax=1
30
+ router_fusion=0
31
+ router_dtype=fp32
32
+ device_initiated_grouped_gemm=0
33
+ norm_epsilon=1e-6
34
+ seq_length=4096
35
+ micro_batch_size=1
36
+ global_batch_size=8
37
+ num_microbatches=1
38
+ assignments_per_rank=32768
39
+ mean_tokens_per_expert=4096
40
+ slot_policy=matched
41
+ slots_per_rank=8
42
+ capacity_mode=dynamic
43
+ capacity=dynamic
44
+ cuda_graph=0
45
+ probe_mode=0
46
+ async_grad_reduce=0
47
+ branch_role=memory
48
+ memory_dump_mode=iteration
49
+ memory_snapshot_path=/workspace/tierb_study_20260826_144644/ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot.pickle
50
+ cuda_graph_warmup_steps=3
51
+ recompute=none
52
+ force_balance=0
53
+ train_iters=5
54
+ ep_size=8
55
+ echo_dispatch_overlap=1
56
+ echo_dispatch_recompute=0
57
+ ultraep_quota_min_tokens=512
58
+ ultraep_weight_sync_join=
59
+ profile_enable=0
60
+ kernel_trace=0
61
+ fp8_recipe=mxfp8
ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/train.log ADDED
@@ -0,0 +1,1221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==============================================================
2
+ QWEN2_57B / ultraep (slots/rank=8 policy=matched)
3
+ E=64 K=8 H'=2560 m=8
4
+ L=10 seq=4096 gbs=8 (1 microbatches)
5
+ cap_mode=dynamic cap=dynamic cg=0 recompute=none
6
+ memory_dump=iteration snapshots=/workspace/tierb_study_20260826_144644/ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot.pickle
7
+ repo=/workspace/Megatron-UltraEP @ c3a8a8a7
8
+ ==============================================================
9
+ W0826 19:28:31.778000 675786 torch/distributed/run.py:766]
10
+ W0826 19:28:31.778000 675786 torch/distributed/run.py:766] *****************************************
11
+ W0826 19:28:31.778000 675786 torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
12
+ W0826 19:28:31.778000 675786 torch/distributed/run.py:766] *****************************************
13
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
14
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
15
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
16
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
17
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
18
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
19
+ WARNING: one_logger package is required to enable e2e metrics tracking. please go to https://confluence.nvidia.com/display/MLWFO/Package+Repositories for details to install it
20
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
21
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
22
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
23
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
24
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
25
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
26
+ using world size: 8, data-parallel size: 8, context-parallel size: 1, hierarchical context-parallel sizes: None, tensor-model-parallel size: 1, pipeline-model-parallel size: 1
27
+ WARNING: overriding default arguments for tokenizer_type:GPT2BPETokenizer with tokenizer_type:GPT2BPETokenizer
28
+ Number of virtual stages per pipeline stage: None
29
+ accumulate and all-reduce gradients in fp32 for bfloat16 data type.
30
+ using torch.bfloat16 for parameters ...
31
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
32
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
33
+ ------------------------ arguments ------------------------
34
+ account_for_embedding_in_pipeline_split ......... False
35
+ account_for_loss_in_pipeline_split .............. False
36
+ accumulate_allreduce_grads_in_fp32 .............. True
37
+ adam_beta1 ...................................... 0.9
38
+ adam_beta2 ...................................... 0.999
39
+ adam_eps ........................................ 1e-08
40
+ add_bias_linear ................................. False
41
+ add_position_embedding .......................... False
42
+ add_qkv_bias .................................... True
43
+ adlr_autoresume ................................. False
44
+ adlr_autoresume_interval ........................ 1000
45
+ align_grad_reduce ............................... True
46
+ align_param_gather .............................. False
47
+ app_tag_run_name ................................ None
48
+ app_tag_run_version ............................. 0.0.0
49
+ apply_layernorm_1p .............................. False
50
+ apply_query_key_layer_scaling ................... False
51
+ apply_residual_connection_post_layernorm ........ False
52
+ apply_rope_fusion ............................... True
53
+ async_save ...................................... None
54
+ async_tensor_model_parallel_allreduce ........... True
55
+ attention_backend ............................... AttnBackend.auto
56
+ attention_dropout ............................... 0.0
57
+ attention_softmax_in_fp32 ....................... False
58
+ auto_detect_ckpt_format ......................... False
59
+ barrier_with_L1_time ............................ True
60
+ bert_binary_head ................................ True
61
+ bert_embedder_type .............................. megatron
62
+ bert_load ....................................... None
63
+ bf16 ............................................ True
64
+ bias_dropout_fusion ............................. True
65
+ bias_gelu_fusion ................................ False
66
+ bias_swiglu_fusion .............................. True
67
+ biencoder_projection_dim ........................ 0
68
+ biencoder_shared_query_context_model ............ False
69
+ block_data_path ................................. None
70
+ calc_ft_timeouts ................................ False
71
+ calculate_per_token_loss ........................ False
72
+ check_for_large_grads ........................... False
73
+ check_for_nan_in_loss_and_grad .................. True
74
+ check_for_spiky_loss ............................ False
75
+ check_weight_hash_across_dp_replicas_interval ... None
76
+ ckpt_assume_constant_structure .................. False
77
+ ckpt_convert_format ............................. None
78
+ ckpt_convert_save ............................... None
79
+ ckpt_convert_update_legacy_dist_opt_format ...... False
80
+ ckpt_format ..................................... torch_dist
81
+ ckpt_fully_parallel_load ........................ False
82
+ ckpt_fully_parallel_save ........................ True
83
+ ckpt_fully_parallel_save_deprecated ............. False
84
+ ckpt_step ....................................... None
85
+ classes_fraction ................................ 1.0
86
+ clip_grad ....................................... 1.0
87
+ clone_scatter_output_in_embedding ............... True
88
+ config_logger_dir ...............................
89
+ consumed_train_samples .......................... 0
90
+ consumed_valid_samples .......................... 0
91
+ context_parallel_size ........................... 1
92
+ cp_comm_type .................................... ['p2p']
93
+ create_attention_mask_in_dataloader ............. True
94
+ cross_entropy_fusion_impl ....................... native
95
+ cross_entropy_loss_fusion ....................... False
96
+ cuda_graph_scope ................................ full
97
+ cuda_graph_warmup_steps ......................... 3
98
+ data_args_path .................................. None
99
+ data_cache_path ................................. None
100
+ data_parallel_random_init ....................... False
101
+ data_parallel_sharding_strategy ................. no_shard
102
+ data_parallel_size .............................. 8
103
+ data_path ....................................... ['/workspace/data/wikitext/wikitext103_gpt2_text_document']
104
+ data_per_class_fraction ......................... 1.0
105
+ data_sharding ................................... True
106
+ dataloader_type ................................. single
107
+ ddp_average_in_collective ....................... False
108
+ ddp_bucket_size ................................. None
109
+ ddp_num_buckets ................................. None
110
+ ddp_pad_buckets_for_high_nccl_busbw ............. False
111
+ decoder_first_pipeline_num_layers ............... None
112
+ decoder_last_pipeline_num_layers ................ None
113
+ decoder_num_layers .............................. None
114
+ decoder_seq_length .............................. None
115
+ decoupled_lr .................................... None
116
+ decoupled_min_lr ................................ None
117
+ decrease_batch_size_if_needed ................... False
118
+ defer_embedding_wgrad_compute ................... False
119
+ delay_wgrad_compute ............................. False
120
+ deprecated_use_mcore_models ..................... True
121
+ deterministic_mode .............................. False
122
+ dino_bottleneck_size ............................ 256
123
+ dino_freeze_last_layer .......................... 1
124
+ dino_head_hidden_size ........................... 2048
125
+ dino_local_crops_number ......................... 10
126
+ dino_local_img_size ............................. 96
127
+ dino_norm_last_layer ............................ False
128
+ dino_teacher_temp ............................... 0.07
129
+ dino_warmup_teacher_temp ........................ 0.04
130
+ dino_warmup_teacher_temp_epochs ................. 30
131
+ disable_bf16_reduced_precision_matmul ........... False
132
+ disable_mamba_mem_eff_path ...................... False
133
+ disable_straggler_on_startup .................... False
134
+ dist_ckpt_format_deprecated ..................... None
135
+ dist_ckpt_strictness ............................ assume_ok_unexpected
136
+ distribute_saved_activations .................... False
137
+ distributed_backend ............................. nccl
138
+ distributed_timeout_minutes ..................... 10
139
+ embedding_init_method_std ....................... None
140
+ embedding_path .................................. None
141
+ empty_unused_memory_level ....................... 0
142
+ enable_cuda_graph ............................... False
143
+ enable_experimental ............................. False
144
+ enable_ft_package ............................... False
145
+ enable_gloo_process_groups ...................... True
146
+ enable_msc ...................................... True
147
+ enable_one_logger ............................... True
148
+ encoder_num_layers .............................. 10
149
+ encoder_seq_length .............................. 4096
150
+ end_weight_decay ................................ 0.1
151
+ eod_mask_loss ................................... False
152
+ error_injection_rate ............................ 0
153
+ error_injection_type ............................ transient_error
154
+ eval_interval ................................... 1000000
155
+ eval_iters ...................................... 0
156
+ evidence_data_path .............................. None
157
+ exit_duration_in_mins ........................... None
158
+ exit_interval ................................... None
159
+ exit_on_missing_checkpoint ...................... False
160
+ exit_signal_handler ............................. False
161
+ exp_avg_dtype ................................... torch.float32
162
+ exp_avg_sq_dtype ................................ torch.float32
163
+ expert_model_parallel_size ...................... 8
164
+ expert_tensor_parallel_size ..................... 1
165
+ export_force_local_attention .................... False
166
+ export_kd_cfg ................................... None
167
+ export_kd_teacher_ckpt_format ................... None
168
+ export_kd_teacher_load .......................... None
169
+ export_kv_cache_quant ........................... False
170
+ export_legacy_megatron .......................... False
171
+ export_model_type ............................... GPTModel
172
+ export_moe_apply_probs_on_input ................. False
173
+ export_qk_l2_norm ............................... False
174
+ export_quant_cfg ................................ None
175
+ export_real_quant_cfg ........................... None
176
+ export_te_mcore_model ........................... False
177
+ external_cuda_graph ............................. False
178
+ ffn_hidden_size ................................. 2560
179
+ finetune ........................................ False
180
+ finetune_data_split ............................. train
181
+ finetune_hf_dataset ............................. None
182
+ first_last_layers_bf16 .......................... False
183
+ flash_decode .................................... False
184
+ fp16 ............................................ False
185
+ fp16_lm_cross_entropy ........................... False
186
+ fp32_residual_connection ........................ False
187
+ fp8 ............................................. e4m3
188
+ fp8_amax_compute_algo ........................... most_recent
189
+ fp8_amax_history_len ............................ 1
190
+ fp8_interval .................................... 1
191
+ fp8_margin ...................................... 0
192
+ fp8_param_gather ................................ False
193
+ fp8_recipe ...................................... mxfp8
194
+ fp8_wgrad ....................................... True
195
+ fsdp_double_buffer .............................. False
196
+ global_batch_size ............................... 8
197
+ grad_reduce_in_bf16 ............................. False
198
+ gradient_accumulation_fusion .................... True
199
+ gradient_reduce_div_fusion ...................... True
200
+ group_query_attention ........................... True
201
+ head_lr_mult .................................... 1.0
202
+ heterogeneous_layers_config_encoded_json ........ None
203
+ heterogeneous_layers_config_path ................ None
204
+ hidden_dropout .................................. 0.0
205
+ hidden_size ..................................... 3584
206
+ hierarchical_context_parallel_sizes ............. None
207
+ high_priority_stream_groups ..................... []
208
+ hybrid_attention_ratio .......................... 0.0
209
+ hybrid_mlp_ratio ................................ 0.0
210
+ hybrid_override_pattern ......................... None
211
+ hysteresis ...................................... 2
212
+ ict_head_size ................................... None
213
+ ict_load ........................................ None
214
+ img_h ........................................... 224
215
+ img_w ........................................... 224
216
+ indexer_batch_size .............................. 128
217
+ indexer_log_interval ............................ 1000
218
+ inference_batch_times_seqlen_threshold .......... -1
219
+ inference_dynamic_batching ...................... False
220
+ inference_dynamic_batching_buffer_guaranteed_fraction 0.2
221
+ inference_dynamic_batching_buffer_overflow_factor None
222
+ inference_dynamic_batching_buffer_size_gb ....... 40.0
223
+ inference_dynamic_batching_chunk_size ........... 256
224
+ inference_dynamic_batching_max_requests_override None
225
+ inference_dynamic_batching_max_tokens_override .. None
226
+ inference_dynamic_batching_num_cuda_graphs ...... 16
227
+ inference_max_batch_size ........................ 8
228
+ inference_max_seq_length ........................ 2560
229
+ inference_rng_tracker ........................... False
230
+ init_method_std ................................. 0.01
231
+ init_method_xavier_uniform ...................... False
232
+ init_model_with_meta_device ..................... False
233
+ initial_loss_scale .............................. 4294967296
234
+ inprocess_active_world_size ..................... 8
235
+ inprocess_barrier_timeout ....................... 120
236
+ inprocess_completion_timeout .................... 120
237
+ inprocess_empty_cuda_cache ...................... False
238
+ inprocess_granularity ........................... node
239
+ inprocess_hard_timeout .......................... 90
240
+ inprocess_heartbeat_interval .................... 30
241
+ inprocess_heartbeat_timeout ..................... 60
242
+ inprocess_last_call_wait ........................ 1
243
+ inprocess_max_iterations ........................ None
244
+ inprocess_monitor_process_interval .............. 1.0
245
+ inprocess_monitor_thread_interval ............... 1.0
246
+ inprocess_progress_watchdog_interval ............ 1.0
247
+ inprocess_restart ............................... False
248
+ inprocess_soft_timeout .......................... 60
249
+ inprocess_termination_grace_time ................ 1
250
+ is_hybrid_model ................................. False
251
+ iter_per_epoch .................................. 1250
252
+ iterations_to_skip .............................. []
253
+ keep_fp8_transpose_cache_when_using_custom_fsdp . False
254
+ kitchen_config_file ............................. None
255
+ kitchen_recipe_number ........................... None
256
+ kv_channels ..................................... 128
257
+ kv_lora_rank .................................... 32
258
+ lazy_mpu_init ................................... None
259
+ load ............................................ None
260
+ load_main_params_from_ckpt ...................... None
261
+ load_model_opt_format ........................... False
262
+ local_rank ...................................... 0
263
+ log_energy ...................................... False
264
+ log_interval .................................... 1
265
+ log_loss_scale_to_tensorboard ................... True
266
+ log_memory_to_tensorboard ....................... False
267
+ log_num_zeros_in_grad ........................... False
268
+ log_params_norm ................................. False
269
+ log_progress .................................... False
270
+ log_straggler ................................... False
271
+ log_throughput .................................. True
272
+ log_timers_to_tensorboard ....................... False
273
+ log_validation_ppl_to_tensorboard ............... False
274
+ log_world_size_to_tensorboard ................... False
275
+ logging_level ................................... None
276
+ loss_scale ...................................... None
277
+ loss_scale_window ............................... 1000
278
+ lr .............................................. 0.0001
279
+ lr_decay_iters .................................. 8
280
+ lr_decay_samples ................................ None
281
+ lr_decay_style .................................. cosine
282
+ lr_warmup_fraction .............................. None
283
+ lr_warmup_init .................................. 0.0
284
+ lr_warmup_iters ................................. 5
285
+ lr_warmup_samples ............................... 0
286
+ lr_wsd_decay_iters .............................. None
287
+ lr_wsd_decay_samples ............................ None
288
+ lr_wsd_decay_style .............................. exponential
289
+ main_grads_dtype ................................ torch.float32
290
+ main_params_dtype ............................... torch.float32
291
+ make_vocab_size_divisible_by .................... 128
292
+ mamba_head_dim .................................. 64
293
+ mamba_num_groups ................................ 8
294
+ mamba_num_heads ................................. None
295
+ mamba_state_dim ................................. 128
296
+ manual_gc ....................................... False
297
+ manual_gc_eval .................................. True
298
+ manual_gc_interval .............................. 0
299
+ mask_factor ..................................... 1.0
300
+ mask_prob ....................................... 0.15
301
+ mask_type ....................................... random
302
+ masked_softmax_fusion ........................... False
303
+ max_position_embeddings ......................... 32768
304
+ max_tokens_to_oom ............................... 12000
305
+ memory_snapshot_path ............................ /workspace/tierb_study_20260826_144644/ultraep/tierbmem/tierbmem_gbs8/QWEN2_57B__ultraep__matched__seq4096__gbs8__capdynamic__nocg__memiteration__it5/rep1/memory_snapshots/snapshot.pickle
306
+ merge_file ...................................... /workspace/data/wikitext/gpt2-merges.txt
307
+ micro_batch_size ................................ 1
308
+ microbatch_group_size_per_vp_stage .............. None
309
+ mid_level_dataset_surplus ....................... 0.005
310
+ min_loss_scale .................................. 1.0
311
+ min_lr .......................................... 1e-05
312
+ mlp_chunks_for_prefill .......................... 1
313
+ mmap_bin_files .................................. True
314
+ mock_data ....................................... False
315
+ moe_apply_probs_on_input ........................ False
316
+ moe_aux_loss_coeff .............................. 0.01
317
+ moe_capacity_monitor ............................ False
318
+ moe_deepep_num_sms .............................. 20
319
+ moe_echo_enable_random_offloading ............... False
320
+ moe_echo_expert_dispatch_overlap ................ False
321
+ moe_echo_expert_dispatcher_type ................. hybridep
322
+ moe_echo_recompute_expert_dispatch .............. False
323
+ moe_enable_deepep ............................... False
324
+ moe_enable_echo ................................. False
325
+ moe_enable_hybridep ............................. True
326
+ moe_enable_ultraep .............................. True
327
+ moe_expert_capacity_factor ...................... None
328
+ moe_extended_tp ................................. False
329
+ moe_ffn_hidden_size ............................. 2560
330
+ moe_grouped_gemm ................................ True
331
+ moe_hybridep_num_sms ............................ 16
332
+ moe_input_jitter_eps ............................ None
333
+ moe_layer_freq .................................. 1
334
+ moe_layer_recompute ............................. False
335
+ moe_num_echo_experts ............................ None
336
+ moe_num_redundant_experts_per_rank .............. 8
337
+ moe_pad_expert_input_to_capacity ................ False
338
+ moe_per_layer_logging ........................... False
339
+ moe_permute_fusion .............................. False
340
+ moe_received_token_capacity ..................... None
341
+ moe_router_bias_update_rate ..................... 0.001
342
+ moe_router_dtype ................................ fp32
343
+ moe_router_enable_expert_bias ................... False
344
+ moe_router_force_load_balancing ................. False
345
+ moe_router_group_topk ........................... None
346
+ moe_router_load_balancing_type .................. aux_loss
347
+ moe_router_num_groups ........................... None
348
+ moe_router_padding_for_fp8 ...................... False
349
+ moe_router_pre_softmax .......................... True
350
+ moe_router_score_function ....................... softmax
351
+ moe_router_topk ................................. 8
352
+ moe_router_topk_scaling_factor .................. None
353
+ moe_shared_expert_intermediate_size ............. 20480
354
+ moe_shared_expert_overlap ....................... False
355
+ moe_token_dispatcher_type ....................... flex
356
+ moe_token_drop_policy ........................... probs
357
+ moe_ultraep_max_microbatches .................... None
358
+ moe_ultraep_overlap_weight_sync ................. True
359
+ moe_ultraep_weight_sync_join .................... before_dispatch
360
+ moe_upcycling_granularity ....................... 1
361
+ moe_use_device_initiated_grouped_gemm ........... False
362
+ moe_use_legacy_grouped_gemm ..................... False
363
+ moe_use_upcycling ............................... False
364
+ moe_z_loss_coeff ................................ None
365
+ mrope_section ................................... None
366
+ mscale .......................................... 1.0
367
+ mscale_all_dim .................................. 0.0
368
+ mtp_loss_scaling_factor ......................... 0.1
369
+ mtp_num_layers .................................. None
370
+ multi_latent_attention .......................... False
371
+ nccl_all_reduce_for_prefill ..................... False
372
+ nccl_communicator_config_path ................... None
373
+ nccl_ub ......................................... False
374
+ no_load_optim ................................... None
375
+ no_load_rng ..................................... None
376
+ no_persist_layer_norm ........................... False
377
+ no_rope_freq .................................... None
378
+ no_save_optim ................................... None
379
+ no_save_rng ..................................... None
380
+ non_persistent_ckpt_type ........................ None
381
+ non_persistent_global_ckpt_dir .................. None
382
+ non_persistent_local_ckpt_algo .................. fully_parallel
383
+ non_persistent_local_ckpt_dir ................... None
384
+ non_persistent_save_interval .................... None
385
+ norm_epsilon .................................... 1e-06
386
+ normalization ................................... RMSNorm
387
+ num_attention_heads ............................. 28
388
+ num_channels .................................... 3
389
+ num_classes ..................................... 1000
390
+ num_dataset_builder_threads ..................... 1
391
+ num_distributed_optimizer_instances ............. 1
392
+ num_experts ..................................... 64
393
+ num_layers ...................................... 10
394
+ num_layers_at_end_in_bf16 ....................... 1
395
+ num_layers_at_start_in_bf16 ..................... 1
396
+ num_layers_per_virtual_pipeline_stage ........... None
397
+ num_query_groups ................................ 4
398
+ num_virtual_stages_per_pipeline_rank ............ None
399
+ num_workers ..................................... 2
400
+ object_storage_cache_path ....................... None
401
+ one_logger_async ................................ False
402
+ one_logger_project .............................. megatron-lm
403
+ one_logger_run_name ............................. None
404
+ onnx_safe ....................................... None
405
+ openai_gelu ..................................... False
406
+ optimizer ....................................... adam
407
+ optimizer_cpu_offload ........................... False
408
+ optimizer_offload_fraction ...................... 1.0
409
+ output_bert_embeddings .......................... False
410
+ overlap_cpu_optimizer_d2h_h2d ................... False
411
+ overlap_grad_reduce ............................. True
412
+ overlap_moe_expert_parallel_comm ................ False
413
+ overlap_p2p_comm ................................ False
414
+ overlap_p2p_comm_warmup_flush ................... False
415
+ overlap_param_gather ............................ True
416
+ overlap_param_gather_with_optimizer_step ........ False
417
+ override_opt_param_scheduler .................... False
418
+ params_dtype .................................... torch.bfloat16
419
+ patch_dim ....................................... 16
420
+ per_split_data_args_path ........................ None
421
+ perform_initialization .......................... True
422
+ pin_cpu_grads ................................... True
423
+ pin_cpu_params .................................. True
424
+ pipeline_model_parallel_comm_backend ............ None
425
+ pipeline_model_parallel_layout .................. None
426
+ pipeline_model_parallel_size .................... 1
427
+ position_embedding_type ......................... rope
428
+ pretrained_checkpoint ........................... None
429
+ profile ......................................... False
430
+ profile_ranks ................................... [0]
431
+ profile_step_end ................................ 12
432
+ profile_step_start .............................. 10
433
+ q_lora_rank ..................................... None
434
+ qk_head_dim ..................................... 128
435
+ qk_l2_norm ...................................... False
436
+ qk_layernorm .................................... False
437
+ qk_pos_emb_head_dim ............................. 64
438
+ query_in_block_prob ............................. 0.1
439
+ rampup_batch_size ............................... None
440
+ rank ............................................ 0
441
+ recompute_granularity ........................... None
442
+ recompute_method ................................ None
443
+ recompute_modules ............................... None
444
+ recompute_num_layers ............................ None
445
+ record_cuda_graph_memory_history ................ False
446
+ record_full_program_memory_history .............. False
447
+ record_iteration_memory_history ................. True
448
+ record_memory_history ........................... False
449
+ relative_attention_max_distance ................. 128
450
+ relative_attention_num_buckets .................. 32
451
+ replication ..................................... False
452
+ replication_factor .............................. 2
453
+ replication_jump ................................ None
454
+ rerun_mode ...................................... validate_results
455
+ reset_attention_mask ............................ False
456
+ reset_position_ids .............................. False
457
+ result_rejected_tracker_filename ................ None
458
+ retriever_report_topk_accuracies ................ []
459
+ retriever_score_scaling ......................... False
460
+ retriever_seq_length ............................ 256
461
+ retro_add_retriever ............................. False
462
+ retro_attention_gate ............................ 1
463
+ retro_cyclic_train_iters ........................ None
464
+ retro_encoder_attention_dropout ................. 0.1
465
+ retro_encoder_hidden_dropout .................... 0.1
466
+ retro_encoder_layers ............................ 2
467
+ retro_num_neighbors ............................. 2
468
+ retro_num_retrieved_chunks ...................... 2
469
+ retro_project_dir ............................... None
470
+ retro_verify_neighbor_count ..................... True
471
+ reuse_grad_buf_for_mxfp8_param_ag ............... False
472
+ rope_scaling_factor ............................. 8.0
473
+ rope_type ....................................... None
474
+ rotary_base ..................................... 1000000
475
+ rotary_interleaved .............................. False
476
+ rotary_percent .................................. 1.0
477
+ rotary_scaling_factor ........................... 1.0
478
+ rotary_seq_len_interpolation_factor ............. None
479
+ run_workload_inspector_server ................... False
480
+ sample_rate ..................................... 1.0
481
+ save ............................................ None
482
+ save_interval ................................... None
483
+ save_retain_interval ............................ None
484
+ scatter_gather_tensors_in_pipeline .............. True
485
+ seed ............................................ 1234
486
+ seq_length ...................................... 4096
487
+ sequence_parallel ............................... False
488
+ sft ............................................. False
489
+ sft_tokenizer_prompt_format ..................... nemotron-h-aligned
490
+ sgd_momentum .................................... 0.9
491
+ short_seq_prob .................................. 0.1
492
+ skip_train ...................................... False
493
+ skipped_train_samples ........................... 0
494
+ spec ............................................ None
495
+ split ........................................... 100,0,0
496
+ squared_relu .................................... False
497
+ start_weight_decay .............................. 0.1
498
+ straggler_ctrlr_port ............................ 65535
499
+ straggler_minmax_count .......................... 1
500
+ suggested_communication_unit_size ............... None
501
+ swiglu .......................................... True
502
+ swin_backbone_type .............................. tiny
503
+ symmetric_ar_type ............................... None
504
+ te_rng_tracker .................................. False
505
+ tensor_model_parallel_size ...................... 1
506
+ tensorboard_dir ................................. None
507
+ tensorboard_log_interval ........................ 1
508
+ tensorboard_queue_size .......................... 1000
509
+ test_data_path .................................. None
510
+ test_mode ....................................... False
511
+ tiktoken_num_special_tokens ..................... 1000
512
+ tiktoken_pattern ................................ None
513
+ tiktoken_special_tokens ......................... None
514
+ timing_log_level ................................ 0
515
+ timing_log_option ............................... minmax
516
+ titles_data_path ................................ None
517
+ tokenizer_model ................................. None
518
+ tokenizer_type .................................. GPT2BPETokenizer
519
+ torch_fsdp2_reshard_after_forward ............... True
520
+ tp_comm_bootstrap_backend ....................... nccl
521
+ tp_comm_bulk_dgrad .............................. True
522
+ tp_comm_bulk_wgrad .............................. True
523
+ tp_comm_overlap ................................. False
524
+ tp_comm_overlap_ag .............................. True
525
+ tp_comm_overlap_cfg ............................. None
526
+ tp_comm_overlap_rs .............................. True
527
+ tp_comm_overlap_rs_dgrad ........................ False
528
+ tp_comm_split_ag ................................ True
529
+ tp_comm_split_rs ................................ True
530
+ train_data_path ................................. None
531
+ train_iters ..................................... 5
532
+ train_samples ................................... None
533
+ train_sync_interval ............................. None
534
+ transformer_impl ................................ transformer_engine
535
+ transformer_pipeline_model_parallel_size ........ 1
536
+ untie_embeddings_and_output_weights ............. True
537
+ use_checkpoint_args ............................. False
538
+ use_checkpoint_opt_param_scheduler .............. False
539
+ use_cpu_initialization .......................... None
540
+ use_custom_fsdp ................................. False
541
+ use_dist_ckpt ................................... True
542
+ use_dist_ckpt_deprecated ........................ False
543
+ use_distributed_optimizer ....................... True
544
+ use_flash_attn .................................. False
545
+ use_legacy_models ............................... False
546
+ use_mp_args_from_checkpoint_args ................ False
547
+ use_one_sent_docs ............................... False
548
+ use_persistent_ckpt_worker ...................... False
549
+ use_precision_aware_optimizer ................... False
550
+ use_pytorch_profiler ............................ False
551
+ use_ring_exchange_p2p ........................... False
552
+ use_rope_scaling ................................ False
553
+ use_rotary_position_embeddings .................. False
554
+ use_sharp ....................................... False
555
+ use_tokenizer_model_from_checkpoint_args ........ True
556
+ use_torch_fsdp2 ................................. False
557
+ use_torch_optimizer_for_cpu_offload ............. False
558
+ use_tp_pp_dp_mapping ............................ False
559
+ v_head_dim ...................................... 128
560
+ valid_data_path ................................. None
561
+ variable_seq_lengths ............................ False
562
+ virtual_pipeline_model_parallel_size ............ None
563
+ vision_backbone_type ............................ vit
564
+ vision_pretraining .............................. False
565
+ vision_pretraining_type ......................... classify
566
+ vocab_extra_ids ................................. 0
567
+ vocab_file ...................................... /workspace/data/wikitext/gpt2-vocab.json
568
+ vocab_size ...................................... None
569
+ wandb_exp_name ..................................
570
+ wandb_project ...................................
571
+ wandb_save_dir ..................................
572
+ weight_decay .................................... 0.1
573
+ weight_decay_incr_style ......................... constant
574
+ wgrad_deferral_limit ............................ 0
575
+ world_size ...................................... 8
576
+ yaml_cfg ........................................ None
577
+ -------------------- end of arguments ---------------------
578
+ INFO:megatron.core.num_microbatches_calculator:setting number of microbatches to constant 1
579
+ > building GPT2BPETokenizer tokenizer ...
580
+ /workspace/Megatron-UltraEP/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
581
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
582
+ > padded vocab (size: 50257) with 47 dummy tokens (new size: 50304)
583
+ WARNING:megatron.core.rerun_state_machine:RerunStateMachine initialized in mode RerunMode.VALIDATE_RESULTS
584
+ > initializing torch distributed ...
585
+ [Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
586
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
587
+ [Gloo] Rank [Gloo] Rank 6 is connected to 37 is connected to peer ranks. 7Expected number of connected peer ranks is : peer ranks. 7Expected number of connected peer ranks is : 7
588
+
589
+ [Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
590
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
591
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
592
+ [Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
593
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
594
+ [Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
595
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
596
+ [Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
597
+ [Gloo] Rank [Gloo] Rank 5 is connected to 67 is connected to peer ranks. 7Expected number of connected peer ranks is : peer ranks. 7Expected number of connected peer ranks is :
598
+ 7
599
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
600
+ [Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
601
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
602
+ > initialized tensor model parallel with size 1
603
+ > initialized pipeline model parallel with size 1
604
+ > setting random seeds to 1234 ...
605
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
606
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
607
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
608
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
609
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
610
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
611
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
612
+ [rank1]:[W826 19:28:50.070809503 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 1] using GPU 1 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
613
+ [rank4]:[W826 19:28:50.070961053 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 4] using GPU 4 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
614
+ [rank5]:[W826 19:28:50.072845500 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 5] using GPU 5 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
615
+ [rank2]:[W826 19:28:50.072911847 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 2] using GPU 2 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
616
+ [rank3]:[W826 19:28:50.074222862 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 3] using GPU 3 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
617
+ [rank6]:[W826 19:28:50.074347358 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 6] using GPU 6 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
618
+ [rank7]:[W826 19:28:50.074432056 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 7] using GPU 7 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
619
+ > compiling dataset index builder ...
620
+ make: Entering directory '/workspace/Megatron-UltraEP/megatron/core/datasets'
621
+ make: Nothing to be done for 'default'.
622
+ make: Leaving directory '/workspace/Megatron-UltraEP/megatron/core/datasets'
623
+ >>> done with dataset index builder. Compilation time: 0.034 seconds
624
+ WARNING: constraints for invoking optimized fused softmax kernel are not met. We default back to unfused kernel invocations.
625
+ > compiling and loading fused kernels ...
626
+ [rank0]:[W826 19:28:50.223895873 ProcessGroupNCCL.cpp:4789] [PG ID 0 PG GUID 0 Rank 0] using GPU 0 as device used by this process is currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect. You can specify device_id in init_process_group() to force use of a particular device.
627
+ >>> done with compiling and loading fused kernels. Compilation time: 3.711 seconds
628
+ time to initialize megatron (seconds): 6.875
629
+ [after megatron is initialized] datetime: 2026-08-26 19:28:55
630
+ building GPT model ...
631
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
632
+ warnings.warn(
633
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
634
+ warnings.warn(
635
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
636
+ warnings.warn(
637
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
638
+ warnings.warn(
639
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
640
+ warnings.warn(
641
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
642
+ warnings.warn(
643
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
644
+ warnings.warn(
645
+ /usr/local/lib/python3.12/dist-packages/transformer_engine/pytorch/cpu_offload.py:708: DeprecationWarning: Offloading weights is deprecated. Using offload_weights=True does not have any effect.
646
+ warnings.warn(
647
+ [INFO] Use MNNVL fabric: False
648
+ [WARN] MAX_NUM_NVL_PEERS is not set. Using detected value 8.
649
+ NVSHMEM v3.7.2
650
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
651
+
652
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
653
+
654
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
655
+
656
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
657
+
658
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
659
+
660
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
661
+
662
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
663
+
664
+ /dvs/p4/build/sw/rel/gpgpu/toolkit/r12.9/main_nvshmem/src/modules/transport/ibrc/ibrc.cpp:1754: error status: 7 (NVSHMEMX_ERROR_INTERNAL) Device enumeration failed.
665
+
666
+ UltraEP: NVL domain size <= 8; forcing weight_sync plan mode to direct (was adaptive).
667
+ > number of parameters on (tensor, pipeline) model parallel rank (0, 0): 7262624256
668
+ INFO:megatron.core.distributed.distributed_data_parallel:Setting up DistributedDataParallel with config DistributedDataParallelConfig(grad_reduce_in_fp32=True, overlap_grad_reduce=True, overlap_param_gather=True, align_param_gather=False, use_distributed_optimizer=True, num_distributed_optimizer_instances=1, check_for_nan_in_grad=True, check_for_large_grads=False, bucket_size=40000000, pad_buckets_for_high_nccl_busbw=False, average_in_collective=False, fp8_param_gather=False, reuse_grad_buf_for_mxfp8_param_ag=False, use_custom_fsdp=False, data_parallel_sharding_strategy='no_shard', gradient_reduce_div_fusion=True, suggested_communication_unit_size=None, preserve_fp32_weights=True, keep_fp8_transpose_cache_when_using_custom_fsdp=False, nccl_ub=False, fsdp_double_buffer=False)
669
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 22
670
+ Params for bucket 1 (180289536 elements, 180289536 padded size):
671
+ module.output_layer.weight
672
+ Params for bucket 2 (73403904 elements, 73403904 padded size):
673
+ module.decoder.final_layernorm.weight
674
+ module.decoder.layers.9.mlp.shared_experts.linear_fc2.weight
675
+ Params for bucket 3 (146800640 elements, 146800640 padded size):
676
+ module.decoder.layers.9.mlp.shared_experts.linear_fc1.weight
677
+ Params for bucket 4 (103001600 elements, 103001600 padded size):
678
+ module.decoder.layers.9.mlp.router.weight
679
+ module.decoder.layers.9.self_attention.linear_qkv.bias
680
+ module.decoder.layers.9.pre_mlp_layernorm.weight
681
+ module.decoder.layers.9.self_attention.linear_proj.weight
682
+ module.decoder.layers.8.mlp.shared_experts.linear_fc2.weight
683
+ module.decoder.layers.9.self_attention.linear_qkv.weight
684
+ module.decoder.layers.9.self_attention.linear_qkv.layer_norm_weight
685
+ Params for bucket 5 (146800640 elements, 146800640 padded size):
686
+ module.decoder.layers.8.mlp.shared_experts.linear_fc1.weight
687
+ Params for bucket 6 (103001600 elements, 103001600 padded size):
688
+ module.decoder.layers.8.self_attention.linear_qkv.weight
689
+ module.decoder.layers.8.self_attention.linear_qkv.layer_norm_weight
690
+ module.decoder.layers.8.mlp.router.weight
691
+ module.decoder.layers.8.self_attention.linear_qkv.bias
692
+ module.decoder.layers.8.pre_mlp_layernorm.weight
693
+ module.decoder.layers.8.self_attention.linear_proj.weight
694
+ module.decoder.layers.7.mlp.shared_experts.linear_fc2.weight
695
+ Params for bucket 7 (146800640 elements, 146800640 padded size):
696
+ module.decoder.layers.7.mlp.shared_experts.linear_fc1.weight
697
+ Params for bucket 8 (103001600 elements, 103001600 padded size):
698
+ module.decoder.layers.7.mlp.router.weight
699
+ module.decoder.layers.7.pre_mlp_layernorm.weight
700
+ module.decoder.layers.7.self_attention.linear_qkv.bias
701
+ module.decoder.layers.7.self_attention.linear_proj.weight
702
+ module.decoder.layers.6.mlp.shared_experts.linear_fc2.weight
703
+ module.decoder.layers.7.self_attention.linear_qkv.weight
704
+ module.decoder.layers.7.self_attention.linear_qkv.layer_norm_weight
705
+ Params for bucket 9 (146800640 elements, 146800640 padded size):
706
+ module.decoder.layers.6.mlp.shared_experts.linear_fc1.weight
707
+ Params for bucket 10 (103001600 elements, 103001600 padded size):
708
+ module.decoder.layers.6.self_attention.linear_qkv.layer_norm_weight
709
+ module.decoder.layers.6.pre_mlp_layernorm.weight
710
+ module.decoder.layers.6.mlp.router.weight
711
+ module.decoder.layers.6.self_attention.linear_qkv.bias
712
+ module.decoder.layers.6.self_attention.linear_qkv.weight
713
+ module.decoder.layers.6.self_attention.linear_proj.weight
714
+ module.decoder.layers.5.mlp.shared_experts.linear_fc2.weight
715
+ Params for bucket 11 (146800640 elements, 146800640 padded size):
716
+ module.decoder.layers.5.mlp.shared_experts.linear_fc1.weight
717
+ Params for bucket 12 (103001600 elements, 103001600 padded size):
718
+ module.decoder.layers.5.self_attention.linear_qkv.weight
719
+ module.decoder.layers.5.self_attention.linear_qkv.layer_norm_weight
720
+ module.decoder.layers.5.mlp.router.weight
721
+ module.decoder.layers.5.pre_mlp_layernorm.weight
722
+ module.decoder.layers.5.self_attention.linear_qkv.bias
723
+ module.decoder.layers.5.self_attention.linear_proj.weight
724
+ module.decoder.layers.4.mlp.shared_experts.linear_fc2.weight
725
+ Params for bucket 13 (146800640 elements, 146800640 padded size):
726
+ module.decoder.layers.4.mlp.shared_experts.linear_fc1.weight
727
+ Params for bucket 14 (103001600 elements, 103001600 padded size):
728
+ module.decoder.layers.4.mlp.router.weight
729
+ module.decoder.layers.4.pre_mlp_layernorm.weight
730
+ module.decoder.layers.4.self_attention.linear_qkv.bias
731
+ module.decoder.layers.4.self_attention.linear_proj.weight
732
+ module.decoder.layers.3.mlp.shared_experts.linear_fc2.weight
733
+ module.decoder.layers.4.self_attention.linear_qkv.weight
734
+ module.decoder.layers.4.self_attention.linear_qkv.layer_norm_weight
735
+ Params for bucket 15 (146800640 elements, 146800640 padded size):
736
+ module.decoder.layers.3.mlp.shared_experts.linear_fc1.weight
737
+ Params for bucket 16 (103001600 elements, 103001600 padded size):
738
+ module.decoder.layers.3.self_attention.linear_qkv.weight
739
+ module.decoder.layers.3.self_attention.linear_qkv.layer_norm_weight
740
+ module.decoder.layers.3.mlp.router.weight
741
+ module.decoder.layers.3.pre_mlp_layernorm.weight
742
+ module.decoder.layers.3.self_attention.linear_qkv.bias
743
+ module.decoder.layers.3.self_attention.linear_proj.weight
744
+ module.decoder.layers.2.mlp.shared_experts.linear_fc2.weight
745
+ Params for bucket 17 (146800640 elements, 146800640 padded size):
746
+ module.decoder.layers.2.mlp.shared_experts.linear_fc1.weight
747
+ Params for bucket 18 (103001600 elements, 103001600 padded size):
748
+ module.decoder.layers.2.mlp.router.weight
749
+ module.decoder.layers.2.self_attention.linear_qkv.bias
750
+ module.decoder.layers.2.pre_mlp_layernorm.weight
751
+ module.decoder.layers.2.self_attention.linear_proj.weight
752
+ module.decoder.layers.1.mlp.shared_experts.linear_fc2.weight
753
+ module.decoder.layers.2.self_attention.linear_qkv.weight
754
+ module.decoder.layers.2.self_attention.linear_qkv.layer_norm_weight
755
+ Params for bucket 19 (146800640 elements, 146800640 padded size):
756
+ module.decoder.layers.1.mlp.shared_experts.linear_fc1.weight
757
+ Params for bucket 20 (103001600 elements, 103001600 padded size):
758
+ module.decoder.layers.1.mlp.router.weight
759
+ module.decoder.layers.1.pre_mlp_layernorm.weight
760
+ module.decoder.layers.1.self_attention.linear_qkv.bias
761
+ module.decoder.layers.1.self_attention.linear_proj.weight
762
+ module.decoder.layers.0.mlp.shared_experts.linear_fc2.weight
763
+ module.decoder.layers.1.self_attention.linear_qkv.layer_norm_weight
764
+ module.decoder.layers.1.self_attention.linear_qkv.weight
765
+ Params for bucket 21 (146800640 elements, 146800640 padded size):
766
+ module.decoder.layers.0.mlp.shared_experts.linear_fc1.weight
767
+ Params for bucket 22 (209890816 elements, 209890816 padded size):
768
+ module.decoder.layers.0.mlp.router.weight
769
+ module.decoder.layers.0.self_attention.linear_qkv.layer_norm_weight
770
+ module.decoder.layers.0.self_attention.linear_qkv.weight
771
+ module.decoder.layers.0.self_attention.linear_proj.weight
772
+ module.embedding.word_embeddings.weight
773
+ module.decoder.layers.0.pre_mlp_layernorm.weight
774
+ module.decoder.layers.0.self_attention.linear_qkv.bias
775
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 44
776
+ Params for bucket 1 (45875200 elements, 45875200 padded size):
777
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight7
778
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight5
779
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight3
780
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight6
781
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight4
782
+ Params for bucket 2 (45875200 elements, 45875200 padded size):
783
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight2
784
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight1
785
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight7
786
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight0
787
+ Params for bucket 3 (55050240 elements, 55050240 padded size):
788
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight6
789
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight5
790
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight4
791
+ Params for bucket 4 (55050240 elements, 55050240 padded size):
792
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight3
793
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight1
794
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight2
795
+ Params for bucket 5 (45875200 elements, 45875200 padded size):
796
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight0
797
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight6
798
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight7
799
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight5
800
+ Params for bucket 6 (45875200 elements, 45875200 padded size):
801
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight4
802
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight1
803
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight3
804
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight0
805
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight2
806
+ Params for bucket 7 (55050240 elements, 55050240 padded size):
807
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight7
808
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight6
809
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight5
810
+ Params for bucket 8 (55050240 elements, 55050240 padded size):
811
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight4
812
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight2
813
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight3
814
+ Params for bucket 9 (45875200 elements, 45875200 padded size):
815
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight1
816
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight0
817
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight7
818
+ Params for bucket 10 (45875200 elements, 45875200 padded size):
819
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight3
820
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight5
821
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight4
822
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight6
823
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight2
824
+ Params for bucket 11 (55050240 elements, 55050240 padded size):
825
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight1
826
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight7
827
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight0
828
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight6
829
+ Params for bucket 12 (55050240 elements, 55050240 padded size):
830
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight5
831
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight4
832
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight3
833
+ Params for bucket 13 (55050240 elements, 55050240 padded size):
834
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight2
835
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight0
836
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight1
837
+ Params for bucket 14 (45875200 elements, 45875200 padded size):
838
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight4
839
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight6
840
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight7
841
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight5
842
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight3
843
+ Params for bucket 15 (45875200 elements, 45875200 padded size):
844
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight2
845
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight0
846
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight1
847
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight7
848
+ Params for bucket 16 (55050240 elements, 55050240 padded size):
849
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight6
850
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight4
851
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight5
852
+ Params for bucket 17 (55050240 elements, 55050240 padded size):
853
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight3
854
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight2
855
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight1
856
+ Params for bucket 18 (45875200 elements, 45875200 padded size):
857
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight0
858
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight5
859
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight7
860
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight6
861
+ Params for bucket 19 (45875200 elements, 45875200 padded size):
862
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight4
863
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight1
864
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight0
865
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight2
866
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight3
867
+ Params for bucket 20 (55050240 elements, 55050240 padded size):
868
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight7
869
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight5
870
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight6
871
+ Params for bucket 21 (55050240 elements, 55050240 padded size):
872
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight4
873
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight2
874
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight3
875
+ Params for bucket 22 (45875200 elements, 45875200 padded size):
876
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight1
877
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight7
878
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight0
879
+ Params for bucket 23 (45875200 elements, 45875200 padded size):
880
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight5
881
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight2
882
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight3
883
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight6
884
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight4
885
+ Params for bucket 24 (55050240 elements, 55050240 padded size):
886
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight1
887
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight0
888
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight7
889
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight6
890
+ Params for bucket 25 (55050240 elements, 55050240 padded size):
891
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight5
892
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight3
893
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight4
894
+ Params for bucket 26 (55050240 elements, 55050240 padded size):
895
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight2
896
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight1
897
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight0
898
+ Params for bucket 27 (45875200 elements, 45875200 padded size):
899
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight7
900
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight3
901
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight5
902
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight6
903
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight4
904
+ Params for bucket 28 (45875200 elements, 45875200 padded size):
905
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight2
906
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight7
907
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight0
908
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight1
909
+ Params for bucket 29 (55050240 elements, 55050240 padded size):
910
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight6
911
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight4
912
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight5
913
+ Params for bucket 30 (55050240 elements, 55050240 padded size):
914
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight3
915
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight2
916
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight1
917
+ Params for bucket 31 (45875200 elements, 45875200 padded size):
918
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight0
919
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight5
920
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight7
921
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight6
922
+ Params for bucket 32 (45875200 elements, 45875200 padded size):
923
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight4
924
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight1
925
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight0
926
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight2
927
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight3
928
+ Params for bucket 33 (55050240 elements, 55050240 padded size):
929
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight7
930
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight5
931
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight6
932
+ Params for bucket 34 (55050240 elements, 55050240 padded size):
933
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight4
934
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight2
935
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight3
936
+ Params for bucket 35 (45875200 elements, 45875200 padded size):
937
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight1
938
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight7
939
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight0
940
+ Params for bucket 36 (45875200 elements, 45875200 padded size):
941
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight6
942
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight2
943
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight4
944
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight5
945
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight3
946
+ Params for bucket 37 (55050240 elements, 55050240 padded size):
947
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight1
948
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight7
949
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight6
950
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight0
951
+ Params for bucket 38 (55050240 elements, 55050240 padded size):
952
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight5
953
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight4
954
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight3
955
+ Params for bucket 39 (55050240 elements, 55050240 padded size):
956
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight2
957
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight0
958
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight1
959
+ Params for bucket 40 (45875200 elements, 45875200 padded size):
960
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight7
961
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight3
962
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight5
963
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight4
964
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight6
965
+ Params for bucket 41 (45875200 elements, 45875200 padded size):
966
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight2
967
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight0
968
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight1
969
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight7
970
+ Params for bucket 42 (55050240 elements, 55050240 padded size):
971
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight6
972
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight4
973
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight5
974
+ Params for bucket 43 (55050240 elements, 55050240 padded size):
975
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight3
976
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight1
977
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight2
978
+ Params for bucket 44 (18350080 elements, 18350080 padded size):
979
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight0
980
+ INFO:megatron.core.optimizer:Setting up optimizer with config OptimizerConfig(optimizer='adam', lr=0.0001, min_lr=1e-05, decoupled_lr=None, decoupled_min_lr=None, weight_decay=0.1, fp8_recipe='mxfp8', fp16=False, bf16=True, reuse_grad_buf_for_mxfp8_param_ag=False, params_dtype=torch.bfloat16, use_precision_aware_optimizer=False, store_param_remainders=True, main_grads_dtype=torch.float32, main_params_dtype=torch.float32, exp_avg_dtype=torch.float32, exp_avg_sq_dtype=torch.float32, loss_scale=None, initial_loss_scale=4294967296, min_loss_scale=1.0, loss_scale_window=1000, hysteresis=2, adam_beta1=0.9, adam_beta2=0.999, adam_eps=1e-08, sgd_momentum=0.9, use_distributed_optimizer=True, overlap_param_gather_with_optimizer_step=False, optimizer_cpu_offload=False, optimizer_offload_fraction=1.0, use_torch_optimizer_for_cpu_offload=False, overlap_cpu_optimizer_d2h_h2d=False, pin_cpu_grads=True, pin_cpu_params=True, clip_grad=1.0, log_num_zeros_in_grad=False, barrier_with_L1_time=True, timers=<megatron.core.timers.Timers object at 0x713c6a33d340>, config_logger_dir='')
981
+ INFO:megatron.core.optimizer_param_scheduler:> learning rate decay style: cosine
982
+ [after model, optimizer, and learning rate scheduler are built] datetime: 2026-08-26 19:28:58
983
+ > building train, validation, and test datasets ...
984
+ > datasets target sizes (minimum size):
985
+ train: 40
986
+ validation: 0
987
+ test: 0
988
+ INFO:megatron.core.datasets.blended_megatron_dataset_config:Let split_matrix = [(0, 1.0), None, None]
989
+ > building train, validation, and test datasets for GPT ...
990
+ INFO:megatron.core.datasets.blended_megatron_dataset_builder:Building GPTDataset splits with sizes=(40, 0, 0) and config=GPTDatasetConfig(random_seed=1234, sequence_length=4096, blend=(['/workspace/data/wikitext/wikitext103_gpt2_text_document'], None), blend_per_split=None, split='100,0,0', split_matrix=[(0, 1.0), None, None], num_dataset_builder_threads=1, path_to_cache=None, mmap_bin_files=True, mock=False, tokenizer=<megatron.training.tokenizer.tokenizer._GPT2BPETokenizer object at 0x713c6a1f2060>, mid_level_dataset_surplus=0.005, reset_position_ids=False, reset_attention_mask=False, eod_mask_loss=False, create_attention_mask=True, drop_last_partial_validation_sequence=True, add_extra_token_to_sequence=True, object_storage_cache_path=None)
991
+ INFO:megatron.core.datasets.indexed_dataset:Load the _IndexReader from /workspace/data/wikitext/wikitext103_gpt2_text_document.idx
992
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence lengths
993
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence pointers
994
+ INFO:megatron.core.datasets.indexed_dataset: Extract the document indices
995
+ INFO:megatron.core.datasets.indexed_dataset:> total number of sequences: 1165029
996
+ INFO:megatron.core.datasets.indexed_dataset:> total number of documents: 1801350
997
+ INFO:megatron.core.datasets.gpt_dataset:Load the GPTDataset train indices
998
+ INFO:megatron.core.datasets.gpt_dataset: Load the document index from 0152a0378ab8ae1561633a6718ca974c-GPTDataset-train-document_index.npy
999
+ INFO:megatron.core.datasets.gpt_dataset: Load the sample index from 0152a0378ab8ae1561633a6718ca974c-GPTDataset-train-sample_index.npy
1000
+ INFO:megatron.core.datasets.gpt_dataset: Load the shuffle index from 0152a0378ab8ae1561633a6718ca974c-GPTDataset-train-shuffle_index.npy
1001
+ INFO:megatron.core.datasets.gpt_dataset:> total number of samples: 29073
1002
+ > finished creating GPT datasets ...
1003
+ [after dataloaders are built] datetime: 2026-08-26 19:28:59
1004
+ done with setup ...
1005
+ (min, max) time across ranks (ms):
1006
+ model-and-optimizer-setup ......................: (3065.75, 3073.56)
1007
+ train/valid/test-data-iterators-setup ..........: (74.39, 91.72)
1008
+ training ...
1009
+ Overwriting rerun_state_machine.current_iteration from -1 to 0...
1010
+ [before the start of training step] datetime: 2026-08-26 19:28:59
1011
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1012
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1013
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1014
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1015
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1016
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1017
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1018
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1019
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1020
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1021
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1022
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1023
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1024
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1025
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1026
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1027
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1028
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1029
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1030
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1031
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1032
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1033
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1034
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1035
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1036
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1037
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1038
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1039
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1040
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1041
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1042
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1043
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1044
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1045
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1046
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1047
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1048
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1049
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1050
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1051
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1052
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1053
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1054
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1055
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1056
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1057
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1058
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1059
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1060
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1061
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1062
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1063
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1064
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1065
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1066
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1067
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1068
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1069
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1070
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1071
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1072
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1073
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1074
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1075
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1076
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1077
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1078
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1079
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1080
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1081
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1082
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1083
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1084
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1085
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1086
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1087
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1088
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1089
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1090
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1091
+ WARNING:megatron.core.rerun_state_machine:Result validation enabled
1092
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1093
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1094
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1095
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1096
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1097
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1098
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1099
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1100
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1101
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1102
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1103
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1104
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1105
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1106
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1107
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1108
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1109
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1110
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1111
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1112
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1113
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1114
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1115
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1116
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1117
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1118
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1119
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1120
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1121
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1122
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1123
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1124
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1125
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1126
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1127
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1128
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1129
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1130
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1131
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1132
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1133
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1134
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1135
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1136
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1137
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1138
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1139
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1140
+ Number of parameters in transformer block in billions: 20.11
1141
+ [2026-08-26 19:29:27] iteration 1/ 5 | consumed samples: 8 | elapsed time per iteration (ms): 28570.8 | throughput per GPU (TFLOP/s/GPU): 4.3 | learning rate: 2.000000E-05 | global batch size: 8 | lm loss: 1.097922E+01 | loss scale: 1.0 | grad norm: 29.874 | number of skipped iterations: 0 | number of nan iterations: 0 |Number of parameters in embedding layers in billions: 0.36
1142
+
1143
+ Total number of parameters in billions: 20.47
1144
+ Number of parameters in most loaded shard in billions: 20.4724
1145
+ Theoretical memory footprints: weight and optimizer=146430.14 MB
1146
+ [Rank 0] (after 1 iterations) memory (MB) | allocated: 64376.66796875 | max allocated: 65220.53515625 | reserved: 73768.0 | max reserved: 73768.0
1147
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1148
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1149
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1150
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1151
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1152
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1153
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1154
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1155
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1156
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1157
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1158
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1159
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1160
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1161
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1162
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1163
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1164
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1165
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1166
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1167
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1168
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1169
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1170
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1171
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1172
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1173
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1174
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1175
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1176
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1177
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1178
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1179
+ [2026-08-26 19:29:28] iteration 2/ 5 | consumed samples: 16 | elapsed time per iteration (ms): 757.0 | throughput per GPU (TFLOP/s/GPU): 163.1 | learning rate: 4.000000E-05 | global batch size: 8 | lm loss: 1.097874E+01 | loss scale: 1.0 | grad norm: 30.624 | number of skipped iterations: 0 | number of nan iterations: 0 |
1180
+ [2026-08-26 19:29:28] iteration 3/ 5 | consumed samples: 24 | elapsed time per iteration (ms): 289.0 | throughput per GPU (TFLOP/s/GPU): 427.3 | learning rate: 6.000000E-05 | global batch size: 8 | lm loss: 9.539532E+00 | loss scale: 1.0 | grad norm: 32.430 | number of skipped iterations: 0 | number of nan iterations: 0 |
1181
+ [2026-08-26 19:29:28] iteration 4/ 5 | consumed samples: 32 | elapsed time per iteration (ms): 285.3 | throughput per GPU (TFLOP/s/GPU): 432.8 | learning rate: 8.000000E-05 | global batch size: 8 | lm loss: 1.109996E+01 | loss scale: 1.0 | grad norm: 107.161 | number of skipped iterations: 0 | number of nan iterations: 0 |
1182
+ [2026-08-26 19:29:29] iteration 5/ 5 | consumed samples: 40 | elapsed time per iteration (ms): 283.5 | throughput per GPU (TFLOP/s/GPU): 435.5 | learning rate: 1.000000E-04 | global batch size: 8 | lm loss: 1.000674E+01 | loss scale: 1.0 | grad norm: 26.659 | number of skipped iterations: 0 | number of nan iterations: 0 |
1183
+ [after training is done] datetime: 2026-08-26 19:29:29
1184
+ [rank0]:[W826 19:29:30.423213688 ProcessGroupNCCL.cpp:1505] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
1185
+ W0826 19:29:30.452000 675786 torch/distributed/elastic/agent/server/api.py:719] Received 15 death signal, shutting down workers
1186
+ W0826 19:29:30.453000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675874 closing signal SIGTERM
1187
+ W0826 19:29:30.453000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675875 closing signal SIGTERM
1188
+ W0826 19:29:30.456000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675876 closing signal SIGTERM
1189
+ W0826 19:29:30.462000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675877 closing signal SIGTERM
1190
+ W0826 19:29:30.467000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675878 closing signal SIGTERM
1191
+ W0826 19:29:30.469000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675879 closing signal SIGTERM
1192
+ W0826 19:29:30.471000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675880 closing signal SIGTERM
1193
+ W0826 19:29:30.473000 675786 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 675881 closing signal SIGTERM
1194
+ Traceback (most recent call last):
1195
+ File "/usr/local/bin/torchrun", line 8, in <module>
1196
+ sys.exit(main())
1197
+ ^^^^^^
1198
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper
1199
+ return f(*args, **kwargs)
1200
+ ^^^^^^^^^^^^^^^^^^
1201
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 892, in main
1202
+ run(args)
1203
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 883, in run
1204
+ elastic_launch(
1205
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 139, in __call__
1206
+ return launch_agent(self._config, self._entrypoint, list(args))
1207
+ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1208
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 261, in launch_agent
1209
+ result = agent.run()
1210
+ ^^^^^^^^^^^
1211
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper
1212
+ result = f(*args, **kwargs)
1213
+ ^^^^^^^^^^^^^^^^^^
1214
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run
1215
+ result = self._invoke_run(role)
1216
+ ^^^^^^^^^^^^^^^^^^^^^^
1217
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run
1218
+ time.sleep(monitor_interval)
1219
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler
1220
+ raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
1221
+ torch.distributed.elastic.multiprocessing.api.SignalException: Process 675786 got signal: 15