source_file
stringclasses
73 values
source_kind
stringclasses
2 values
gpu_name
stringclasses
10 values
gpu_uuid
stringclasses
11 values
seed
int64
11
22
model_key
stringclasses
9 values
model_label
stringclasses
9 values
actual_unique_parameters
float64
150M
7.98B
⌀
sequence_length
int64
512
32.8k
batch_size
float64
1
512
⌀
tokens_per_step
float64
1.02k
131k
⌀
status
stringclasses
3 values
stable
bool
1 class
warmup_steps_executed
float64
11
45
⌀
measured_steps
float64
20
20
⌀
median_step_time_ms
float64
42.4
3.59k
⌀
mean_based_tokens_per_second
float64
3.64k
244k
⌀
measured_robust_relative_jitter
float64
0
0.06
⌀
tokens_per_second
float64
3.64k
244k
⌀
achieved_tflops
float64
19.8
668
⌀
dense_bf16_peak_tflops
float64
155
2.25k
model_flops_utilization_pct
float64
0.88
100
⌀
configured_clock_dense_bf16_peak_tflops
float64
181
2.25k
configured_clock_model_flops_utilization_pct
float64
0.88
78
⌀
peak_allocated_gb
float64
4.62
286
⌀
peak_allocated_vram_pct
float64
1.61
99.6
⌀
peak_reserved_gb
float64
4.66
287
⌀
peak_reserved_vram_pct
float64
1.62
99.7
⌀
selected_sdpa_backend
stringclasses
1 value
error
stringclasses
553 values
rtx6000_ada_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX 6000 Ada Generation
GPU-5b377adf-91ee-db42-9c14-0b8ba36df6b5
11
150m
150M
150,436,608
1,024
8
8,192
complete
true
11
20
138.682144
59,061.626572
0.001529
59,070.329849
64.467192
364.2
17.701041
451.433533
14.28055
12.094076
23.77128
12.68777
24.938202
FLASH_ATTENTION
null
rtx6000_ada_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX 6000 Ada Generation
GPU-5b377adf-91ee-db42-9c14-0b8ba36df6b5
11
150m
150M
150,436,608
1,024
16
16,384
complete
true
12
20
285.771957
57,273.277212
0.004257
57,332.427398
62.570509
364.2
17.180261
451.433533
13.860403
22.062327
43.364184
23.286776
45.770875
FLASH_ATTENTION
null
a6000_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX A6000
GPU-18d70bd6-a015-e960-7fda-6a87d5e18b1f
11
150m
150M
150,436,608
1,024
16
16,384
complete
true
11
20
454.921799
35,994.748746
0.001649
36,014.981139
39.30543
154.8
25.391104
180.6
21.763804
22.062327
43.230703
23.244833
45.547799
FLASH_ATTENTION
null
a6000_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX A6000
GPU-18d70bd6-a015-e960-7fda-6a87d5e18b1f
11
150m
150M
150,436,608
1,024
32
32,768
complete
true
11
20
878.285431
37,305.001935
0.000599
37,309.055629
40.717735
154.8
26.303446
180.6
22.545811
41.998829
82.295891
44.413485
87.027363
FLASH_ATTENTION
null
a6000_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX A6000
GPU-18d70bd6-a015-e960-7fda-6a87d5e18b1f
11
150m
150M
150,436,608
1,024
64
null
oom
null
null
null
null
null
null
null
null
154.8
null
180.6
null
50.399771
98.757374
50.625249
99.199195
null
CUDA out of memory. Tried to allocate 256.00 MiB. GPU 0 has a total capacity of 47.53 GiB of which 174.25 MiB is free. Process 1415479 has 47.35 GiB memory in use. Of the allocated memory 46.94 GiB is allocated by PyTorch, and 95.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large...
a6000_context_frontier_seed22_v2.json
context_frontier
NVIDIA RTX A6000
GPU-18d70bd6-a015-e960-7fda-6a87d5e18b1f
22
150m
150M
150,436,608
1,024
16
16,384
complete
true
11
20
453.889496
36,086.451273
0.000695
36,096.891754
39.394824
154.8
25.448853
180.6
21.813302
22.062327
43.230703
23.244833
45.547799
FLASH_ATTENTION
null
a6000_context_frontier_seed22_v2.json
context_frontier
NVIDIA RTX A6000
GPU-18d70bd6-a015-e960-7fda-6a87d5e18b1f
22
150m
150M
150,436,608
1,024
32
32,768
complete
true
11
20
877.787994
37,327.238673
0.000592
37,330.198419
40.74081
154.8
26.318352
180.6
22.558588
41.998829
82.295891
44.413485
87.027363
FLASH_ATTENTION
null
a6000_context_frontier_seed22_v2.json
context_frontier
NVIDIA RTX A6000
GPU-18d70bd6-a015-e960-7fda-6a87d5e18b1f
22
150m
150M
150,436,608
1,024
64
null
oom
null
null
null
null
null
null
null
null
154.8
null
180.6
null
50.399771
98.757374
50.625249
99.199195
null
CUDA out of memory. Tried to allocate 256.00 MiB. GPU 0 has a total capacity of 47.53 GiB of which 174.25 MiB is free. Process 1449495 has 47.35 GiB memory in use. Of the allocated memory 46.94 GiB is allocated by PyTorch, and 95.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large...
rtxpro6000_server_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU-0adc4b73-c849-be4e-7fc3-492d46abdd3e
11
150m
150M
150,436,608
1,024
8
8,192
complete
true
11
20
85.922222
95,345.271789
0.00035
95,342.040699
104.052807
468
22.233506
468
22.233506
12.094076
11.859967
12.685672
12.440111
FLASH_ATTENTION
null
rtxpro6000_server_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU-0adc4b73-c849-be4e-7fc3-492d46abdd3e
11
150m
150M
150,436,608
1,024
16
16,384
complete
true
11
20
155.339333
105,433.994469
0.000348
105,472.321323
115.108624
468
24.59586
468
24.59586
22.062327
21.635258
23.244833
22.794873
FLASH_ATTENTION
null
rtxpro6000_server_context_frontier_seed11_v2.json
context_frontier
NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU-0adc4b73-c849-be4e-7fc3-492d46abdd3e
11
150m
150M
150,436,608
1,024
32
32,768
complete
true
11
20
341.523651
95,941.623159
0.000114
95,946.50295
104.712495
468
22.374465
468
22.374465
41.998829
41.185841
44.325405
43.46738
FLASH_ATTENTION
null
rtxpro6000_server_context_frontier_seed22_v2.json
context_frontier
NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU-0adc4b73-c849-be4e-7fc3-492d46abdd3e
22
150m
150M
150,436,608
1,024
8
8,192
complete
true
11
20
86.09544
95,144.684711
0.000667
95,150.219436
103.84346
468
22.188774
468
22.188774
12.114606
11.880098
13.679722
13.414919
FLASH_ATTENTION
null
rtxpro6000_server_context_frontier_seed22_v2.json
context_frontier
NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU-0adc4b73-c849-be4e-7fc3-492d46abdd3e
22
150m
150M
150,436,608
1,024
16
16,384
complete
true
11
20
155.770126
105,121.851594
0.000396
105,180.629846
114.790282
468
24.527838
468
24.527838
22.078687
21.651301
25.409094
24.91724
FLASH_ATTENTION
null
rtxpro6000_server_context_frontier_seed22_v2.json
context_frontier
NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU-0adc4b73-c849-be4e-7fc3-492d46abdd3e
22
150m
150M
150,436,608
1,024
32
32,768
complete
true
11
20
341.920181
95,831.647606
0.000208
95,835.232299
104.591058
468
22.348517
468
22.348517
42.016786
41.20345
44.421874
43.561982
FLASH_ATTENTION
null
a100_pcie_training_seed11_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
11
150m
150M
150,436,608
2,048
1
2,048
complete
true
13
20
62.858273
32,197.716614
0.009749
32,581.232618
41.707464
312
13.367777
312
13.367777
4.620813
5.437899
4.657775
5.481396
FLASH_ATTENTION
null
a100_pcie_training_seed11_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
11
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
76.162434
53,671.078702
0.004135
53,779.79412
68.843891
312
22.06535
312
22.06535
7.113212
8.371022
7.43021
8.744073
FLASH_ATTENTION
null
a100_pcie_training_seed11_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
11
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
133.296623
61,408.609796
0.003477
61,456.920674
78.671435
312
25.215203
312
25.215203
12.097978
14.23723
12.689867
14.933781
FLASH_ATTENTION
null
a100_pcie_training_seed11_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
11
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
248.006561
65,917.761565
0.002005
66,062.768321
84.567413
312
27.10494
312
27.10494
22.067509
25.969645
23.29097
27.409448
FLASH_ATTENTION
null
a100_pcie_training_seed11_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
11
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
476.017868
68,807.783434
0.00252
68,837.75211
88.119689
312
28.24349
312
28.24349
42.00657
49.434476
44.417679
52.271935
FLASH_ATTENTION
null
a100_pcie_training_seed11_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
11
150m
150M
150,436,608
2,048
32
65,536
complete
true
22
20
1,261.249634
51,677.225658
0.022289
51,961.16474
66.515851
312
21.319183
312
21.319183
81.884694
96.364138
82.250301
96.794395
FLASH_ATTENTION
null
a100_pcie_training_seed11_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
11
150m
150M
150,436,608
2,048
64
null
oom
null
null
null
null
null
null
null
null
312
null
312
null
83.967307
98.815014
84.334871
99.247573
null
CUDA out of memory. Tried to allocate 384.00 MiB. GPU 0 has a total capacity of 79.14 GiB of which 324.75 MiB is free. Process 703773 has 78.81 GiB memory in use. Of the allocated memory 78.20 GiB is allocated by PyTorch, and 110.54 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large...
a100_pcie_training_seed22_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
22
150m
150M
150,436,608
2,048
1
2,048
complete
true
12
20
66.795471
30,895.521427
0.036187
30,660.761328
39.249055
312
12.579825
312
12.579825
4.620813
5.437899
4.657775
5.481396
FLASH_ATTENTION
null
a100_pcie_training_seed22_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
22
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
76.813774
53,048.436301
0.003098
53,323.769695
68.260131
312
21.878247
312
21.878247
7.113212
8.371022
7.43021
8.744073
FLASH_ATTENTION
null
a100_pcie_training_seed22_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
22
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
133.851967
61,098.934515
0.006445
61,201.939667
78.345032
312
25.110587
312
25.110587
12.097978
14.23723
12.689867
14.933781
FLASH_ATTENTION
null
a100_pcie_training_seed22_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
22
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
249.122955
65,764.963144
0.003864
65,766.721412
84.188441
312
26.983475
312
26.983475
22.067509
25.969645
23.29097
27.409448
FLASH_ATTENTION
null
a100_pcie_training_seed22_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
22
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
477.129044
68,701.776081
0.000741
68,677.437354
87.914469
312
28.177714
312
28.177714
42.00657
49.434476
44.417679
52.271935
FLASH_ATTENTION
null
a100_pcie_training_seed22_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
22
150m
150M
150,436,608
2,048
32
65,536
complete
true
21
20
1,271.076233
49,939.454038
0.030666
51,559.456705
66.001622
312
21.154366
312
21.154366
81.884694
96.364138
82.250301
96.794395
FLASH_ATTENTION
null
a100_pcie_training_seed22_v2.json
baseline_reused
NVIDIA A100 80GB PCIe
GPU-2e75422d-0492-4288-c66f-44c400810994
22
150m
150M
150,436,608
2,048
64
null
oom
null
null
null
null
null
null
null
null
312
null
312
null
83.967307
98.815014
84.334871
99.247573
null
CUDA out of memory. Tried to allocate 384.00 MiB. GPU 0 has a total capacity of 79.14 GiB of which 324.75 MiB is free. Process 736416 has 78.81 GiB memory in use. Of the allocated memory 78.20 GiB is allocated by PyTorch, and 110.54 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large...
b200_training_seed11_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
11
150m
150M
150,436,608
2,048
1
2,048
complete
true
27
20
125.284451
16,392.029633
0.018644
16,346.80115
20.925655
2,250
0.930029
2,250
0.930029
4.620813
2.412919
4.657775
2.43222
FLASH_ATTENTION
null
b200_training_seed11_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
11
150m
150M
150,436,608
2,048
2
4,096
complete
true
17
20
125.831425
32,325.355806
0.031547
32,551.487113
41.669387
2,250
1.851973
2,250
1.851973
7.113212
3.714413
7.43021
3.879944
FLASH_ATTENTION
null
b200_training_seed11_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
11
150m
150M
150,436,608
2,048
4
8,192
complete
true
14
20
132.893715
61,229.123497
0.017928
61,643.246303
78.909952
2,250
3.507109
2,250
3.507109
12.097978
6.317383
12.689867
6.626458
FLASH_ATTENTION
null
b200_training_seed11_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
11
150m
150M
150,436,608
2,048
8
16,384
complete
true
25
20
130.838036
124,192.325787
0.017533
125,223.524848
160.299512
2,250
7.124423
2,250
7.124423
22.067509
11.523322
23.29097
12.162195
FLASH_ATTENTION
null
b200_training_seed11_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
11
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
176.223953
185,931.51996
0.004358
185,945.210037
238.029766
2,250
10.579101
2,250
10.579101
42.00657
21.935201
44.417679
23.194246
FLASH_ATTENTION
null
b200_training_seed11_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
11
150m
150M
150,436,608
2,048
32
65,536
complete
true
11
20
319.716721
204,711.119644
0.00129
204,981.459465
262.398203
2,250
11.662142
2,250
11.662142
81.884694
42.75896
86.50752
45.17293
FLASH_ATTENTION
null
b200_training_seed11_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
11
150m
150M
150,436,608
2,048
64
131,072
complete
true
11
20
613.649231
213,340.605878
0.000857
213,594.335962
273.423606
2,250
12.15216
2,250
12.15216
161.640941
84.406476
170.79835
89.188338
FLASH_ATTENTION
null
b200_training_seed22_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
22
150m
150M
150,436,608
2,048
1
2,048
complete
true
11
20
132.093773
15,393.261892
0.019169
15,504.137366
19.846955
2,250
0.882087
2,250
0.882087
4.620813
2.412919
4.657775
2.43222
FLASH_ATTENTION
null
b200_training_seed22_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
22
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
134.421646
30,361.796066
0.018807
30,471.28285
39.006503
2,250
1.733622
2,250
1.733622
7.113212
3.714413
7.43021
3.879944
FLASH_ATTENTION
null
b200_training_seed22_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
22
150m
150M
150,436,608
2,048
4
8,192
complete
true
23
20
140.115791
58,061.099137
0.040921
58,465.929663
74.842646
2,250
3.32634
2,250
3.32634
12.097978
6.317383
12.689867
6.626458
FLASH_ATTENTION
null
b200_training_seed22_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
22
150m
150M
150,436,608
2,048
8
16,384
complete
true
16
20
134.892784
120,454.744326
0.011685
121,459.425032
155.481061
2,250
6.910269
2,250
6.910269
22.067509
11.523322
23.29097
12.162195
FLASH_ATTENTION
null
b200_training_seed22_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
22
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
176.619598
184,074.513656
0.003101
185,528.674615
237.496557
2,250
10.555403
2,250
10.555403
42.00657
21.935201
44.417679
23.194246
FLASH_ATTENTION
null
b200_training_seed22_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
22
150m
150M
150,436,608
2,048
32
65,536
complete
true
11
20
319.687653
204,856.370981
0.001715
205,000.097656
262.422062
2,250
11.663203
2,250
11.663203
81.884694
42.75896
86.50752
45.17293
FLASH_ATTENTION
null
b200_training_seed22_v2.json
baseline_reused
NVIDIA B200
GPU-c4a7ec32-9030-147d-7e07-b6668d499a32
22
150m
150M
150,436,608
2,048
64
131,072
complete
true
11
20
613.760101
213,522.740002
0.001013
213,555.752025
273.374215
2,250
12.149965
2,250
12.149965
161.640941
84.406476
170.79835
89.188338
FLASH_ATTENTION
null
b300_training_seed11_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
11
150m
150M
150,436,608
2,048
1
2,048
complete
true
11
20
49.818592
41,104.086645
0.001803
41,109.150517
52.624112
2,250
2.338849
2,250
2.338849
4.620813
1.607638
4.657775
1.620498
FLASH_ATTENTION
null
b300_training_seed11_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
11
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
54.790895
74,756.98219
0.003519
74,756.945756
95.69689
2,250
4.253195
2,250
4.253195
7.113212
2.474775
7.43021
2.585062
FLASH_ATTENTION
null
b300_training_seed11_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
11
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
62.252958
129,045.201525
0.007216
131,592.13994
168.452021
2,250
7.486756
2,250
7.486756
12.097978
4.209037
12.689867
4.414963
FLASH_ATTENTION
null
b300_training_seed11_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
11
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
97.369518
168,198.22911
0.000973
168,266.211946
215.398757
2,250
9.573278
2,250
9.573278
22.067509
7.677561
23.29097
8.103218
FLASH_ATTENTION
null
b300_training_seed11_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
11
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
171.559616
190,346.14424
0.00045
191,000.660569
244.501284
2,250
10.866724
2,250
10.866724
42.00657
14.614608
44.417679
15.453463
FLASH_ATTENTION
null
b300_training_seed11_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
11
150m
150M
150,436,608
2,048
32
65,536
complete
true
11
20
315.691162
207,553.380538
0.000394
207,595.295231
265.744192
2,250
11.810853
2,250
11.810853
81.884694
28.488704
86.50752
30.097042
FLASH_ATTENTION
null
b300_training_seed11_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
11
150m
150M
150,436,608
2,048
64
131,072
complete
true
11
20
602.686676
217,469.357714
0.000185
217,479.505046
278.397039
2,250
12.373202
2,250
12.373202
161.640941
56.236894
170.79835
59.422871
FLASH_ATTENTION
null
b300_training_seed22_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
22
150m
150M
150,436,608
2,048
1
2,048
complete
true
11
20
54.70768
37,461.834046
0.008076
37,435.329179
47.921227
2,250
2.129832
2,250
2.129832
4.620813
1.607638
4.657775
1.620498
FLASH_ATTENTION
null
b300_training_seed22_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
22
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
55.767887
73,416.404825
0.010816
73,447.286814
94.020387
2,250
4.178684
2,250
4.178684
7.113212
2.474775
7.43021
2.585062
FLASH_ATTENTION
null
b300_training_seed22_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
22
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
62.259199
131,604.780072
0.0016
131,578.949181
168.435135
2,250
7.486006
2,250
7.486006
12.097978
4.209037
12.689867
4.414963
FLASH_ATTENTION
null
b300_training_seed22_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
22
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
97.440449
168,151.509806
0.001505
168,143.724791
215.24196
2,250
9.566309
2,250
9.566309
22.067509
7.677561
23.29097
8.103218
FLASH_ATTENTION
null
b300_training_seed22_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
22
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
171.624222
190,899.356831
0.000475
190,928.760847
244.409245
2,250
10.862633
2,250
10.862633
42.00657
14.614608
44.417679
15.453463
FLASH_ATTENTION
null
b300_training_seed22_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
22
150m
150M
150,436,608
2,048
32
65,536
complete
true
11
20
315.790024
207,527.076257
0.000246
207,530.30514
265.660998
2,250
11.807155
2,250
11.807155
81.884694
28.488704
86.50752
30.097042
FLASH_ATTENTION
null
b300_training_seed22_v2.json
baseline_reused
NVIDIA B300 SXM6 AC
GPU-589e52a7-94b4-3cdf-33cc-048fbf864c10
22
150m
150M
150,436,608
2,048
64
131,072
complete
true
11
20
602.943848
217,226.369176
0.00014
217,386.744237
278.278295
2,250
12.367924
2,250
12.367924
161.640941
56.236894
170.79835
59.422871
FLASH_ATTENTION
null
rtx4090_training_seed11_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
11
150m
150M
150,436,608
2,048
1
2,048
complete
true
16
20
46.642368
43,738.768099
0.003146
43,908.576556
56.207677
165.2
34.024017
205.516667
27.349449
4.620813
18.292247
4.657775
18.438564
FLASH_ATTENTION
null
rtx4090_training_seed11_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
11
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
73.94429
55,379.250439
0.001833
55,393.053217
70.909036
165.2
42.923145
205.516667
34.502815
7.113212
28.158818
7.43021
29.413703
FLASH_ATTENTION
null
rtx4090_training_seed11_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
11
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
140.867706
58,160.768356
0.001157
58,153.85382
74.443156
165.2
45.062443
205.516667
36.222442
12.097978
47.891829
12.689867
50.234918
FLASH_ATTENTION
null
rtx4090_training_seed11_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
11
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
299.636993
54,677.853634
0.000306
54,679.496726
69.995607
165.2
42.370222
205.516667
34.05836
22.067509
87.357852
23.29097
92.201124
FLASH_ATTENTION
null
rtx4090_training_seed11_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
11
150m
150M
150,436,608
2,048
16
null
oom
null
null
null
null
null
null
null
null
165.2
null
205.516667
null
24.593266
97.356475
24.733811
97.912845
null
CUDA out of memory. Tried to allocate 256.00 MiB. GPU 0 has a total capacity of 23.53 GiB of which 23.69 MiB is free. Including non-PyTorch memory, this process has 23.49 GiB memory in use. Of the allocated memory 22.90 GiB is allocated by PyTorch, and 134.03 MiB is reserved by PyTorch but unallocated. If reserved but ...
rtx4090_training_seed22_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
22
150m
150M
150,436,608
2,048
1
2,048
complete
true
13
20
46.608864
43,949.30518
0.000599
43,940.139958
56.248081
165.2
34.048475
205.516667
27.369109
4.620813
18.292247
4.657775
18.438564
FLASH_ATTENTION
null
rtx4090_training_seed22_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
22
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
74.003231
55,346.864236
0.001981
55,348.934661
70.852559
165.2
42.888958
205.516667
34.475335
7.113212
28.158818
7.43021
29.413703
FLASH_ATTENTION
null
rtx4090_training_seed22_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
22
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
140.984833
57,966.62042
0.001352
58,105.541138
74.38131
165.2
45.025006
205.516667
36.19235
12.097978
47.891829
12.689867
50.234918
FLASH_ATTENTION
null
rtx4090_training_seed22_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
22
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
300.800171
54,435.818596
0.004594
54,468.054161
69.724938
165.2
42.206379
205.516667
33.926659
22.067509
87.357852
23.29097
92.201124
FLASH_ATTENTION
null
rtx4090_training_seed22_v2.json
baseline_reused
NVIDIA GeForce RTX 4090
GPU-5447a6b8-ccd6-9d49-4872-492ff5116ee3
22
150m
150M
150,436,608
2,048
16
null
oom
null
null
null
null
null
null
null
null
165.2
null
205.516667
null
24.593266
97.356475
24.733811
97.912845
null
CUDA out of memory. Tried to allocate 256.00 MiB. GPU 0 has a total capacity of 23.53 GiB of which 23.69 MiB is free. Including non-PyTorch memory, this process has 23.49 GiB memory in use. Of the allocated memory 22.90 GiB is allocated by PyTorch, and 134.03 MiB is reserved by PyTorch but unallocated. If reserved but ...
rtx5090_training_seed11.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
11
150m
150M
150,436,608
2,048
1
2,048
complete
true
11
20
42.378672
48,317.229762
0.000877
48,326.196184
61.862702
209.5
29.528736
268.946822
23.001834
4.620813
13.724295
4.657775
13.834074
null
null
rtx5090_training_seed11.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
11
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
62.230192
65,817.93085
0.000539
65,820.140614
84.256823
209.5
40.218054
268.946822
31.328432
7.113212
21.126979
7.43021
22.068493
null
null
rtx5090_training_seed11.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
11
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
102.0952
80,231.124485
0.000415
80,238.836236
102.714297
209.5
49.028304
268.946822
38.191303
12.097978
35.932249
12.689867
37.69022
null
null
rtx5090_training_seed11.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
11
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
188.242096
86,996.654912
0.001201
87,036.854948
111.416488
209.5
53.182094
268.946822
41.426958
22.067509
65.54279
23.29097
69.176597
null
null
rtx5090_training_seed11.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
11
150m
150M
150,436,608
2,048
16
null
oom
null
null
null
null
null
null
null
null
209.5
null
268.946822
null
31.62633
93.933481
31.822184
94.515188
null
CUDA out of memory. Tried to allocate 1.95 GiB. GPU 0 has a total capacity of 31.36 GiB of which 1.24 GiB is free. Including non-PyTorch memory, this process has 30.11 GiB memory in use. Of the allocated memory 29.36 GiB is allocated by PyTorch, and 162.78 MiB is reserved by PyTorch but unallocated. If reserved but una...
rtx5090_training_seed22.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
22
150m
150M
150,436,608
2,048
1
2,048
complete
true
11
20
42.555616
48,116.827642
0.000839
48,125.257587
61.605479
209.5
29.405957
268.946822
22.906194
4.620813
13.724295
4.657775
13.834074
null
null
rtx5090_training_seed22.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
22
150m
150M
150,436,608
2,048
2
4,096
complete
true
11
20
62.426384
65,603.962806
0.000808
65,613.283028
83.992023
209.5
40.091658
268.946822
31.229974
7.113212
21.126979
7.43021
22.068493
null
null
rtx5090_training_seed22.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
22
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
102.392673
79,983.003473
0.00084
80,005.724244
102.415888
209.5
48.885866
268.946822
38.080349
12.097978
35.932249
12.689867
37.69022
null
null
rtx5090_training_seed22.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
22
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
188.343842
86,993.162099
0.000133
86,989.836593
111.356299
209.5
53.153365
268.946822
41.404579
22.067509
65.54279
23.29097
69.176597
null
null
rtx5090_training_seed22.json
baseline_reused
NVIDIA GeForce RTX 5090
GPU-c3733010-4951-8bd3-2692-36d9198c8d52
22
150m
150M
150,436,608
2,048
16
null
oom
null
null
null
null
null
null
null
null
209.5
null
268.946822
null
31.62633
93.933481
31.822184
94.515188
null
CUDA out of memory. Tried to allocate 1.95 GiB. GPU 0 has a total capacity of 31.36 GiB of which 1.24 GiB is free. Including non-PyTorch memory, this process has 30.11 GiB memory in use. Of the allocated memory 29.36 GiB is allocated by PyTorch, and 162.78 MiB is reserved by PyTorch but unallocated. If reserved but una...
h100_sxm_training_seed11_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
11
150m
150M
150,436,608
2,048
1
2,048
complete
true
24
20
49.361103
41,303.199869
0.011509
41,490.158711
53.111843
989.5
5.367543
989.5
5.367543
4.670883
5.494026
4.720689
5.552609
FLASH_ATTENTION
null
h100_sxm_training_seed11_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
11
150m
150M
150,436,608
2,048
2
4,096
complete
true
14
20
50.452576
80,988.366852
0.004296
81,185.151491
103.925681
989.5
10.502848
989.5
10.502848
7.163282
8.425656
7.493124
8.813626
FLASH_ATTENTION
null
h100_sxm_training_seed11_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
11
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
71.21603
115,025.017985
0.001541
115,030.281611
147.25107
989.5
14.881361
989.5
14.881361
12.148047
14.28888
12.73181
14.975518
FLASH_ATTENTION
null
h100_sxm_training_seed11_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
11
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
126.726883
129,265.066641
0.00088
129,285.906988
165.499796
989.5
16.725598
989.5
16.725598
22.117578
26.015326
23.353885
27.469505
FLASH_ATTENTION
null
h100_sxm_training_seed11_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
11
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
234.329056
139,751.536406
0.000259
139,837.545498
179.007022
989.5
18.090654
989.5
18.090654
42.05664
49.468219
44.459622
52.294676
FLASH_ATTENTION
null
h100_sxm_training_seed11_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
11
150m
150M
150,436,608
2,048
32
65,536
complete
true
11
20
514.424438
126,780.759908
0.002957
127,396.746924
163.081469
989.5
16.481199
989.5
16.481199
81.934764
96.374005
82.397102
96.917821
FLASH_ATTENTION
null
h100_sxm_training_seed11_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
11
150m
150M
150,436,608
2,048
64
null
oom
null
null
null
null
null
null
null
null
989.5
null
989.5
null
84.017376
98.823634
84.18807
99.024408
null
CUDA out of memory. Tried to allocate 384.00 MiB. GPU 0 has a total capacity of 79.18 GiB of which 120.19 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 78.25 GiB is allocated by PyTorch, and 82.79 MiB is reserved by PyTorch but unallocated. If reserved but ...
h100_sxm_training_seed22_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
22
150m
150M
150,436,608
2,048
1
2,048
complete
true
24
20
50.385777
40,601.166223
0.002774
40,646.391531
52.031731
989.5
5.258386
989.5
5.258386
4.670883
5.494026
4.720689
5.552609
FLASH_ATTENTION
null
h100_sxm_training_seed22_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
22
150m
150M
150,436,608
2,048
2
4,096
complete
true
16
20
50.431599
80,557.55692
0.006106
81,218.92045
103.968909
989.5
10.507217
989.5
10.507217
7.163282
8.425656
7.493124
8.813626
FLASH_ATTENTION
null
h100_sxm_training_seed22_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
22
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
71.350044
114,820.427184
0.001261
114,814.224519
146.974493
989.5
14.85341
989.5
14.85341
12.148047
14.28888
12.73181
14.975518
FLASH_ATTENTION
null
h100_sxm_training_seed22_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
22
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
126.941502
128,996.996236
0.00111
129,067.324644
165.219987
989.5
16.697321
989.5
16.697321
22.117578
26.015326
23.353885
27.469505
FLASH_ATTENTION
null
h100_sxm_training_seed22_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
22
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
234.770607
139,571.951436
0.000292
139,574.542229
178.67035
989.5
18.05663
989.5
18.05663
42.05664
49.468219
44.459622
52.294676
FLASH_ATTENTION
null
h100_sxm_training_seed22_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
22
150m
150M
150,436,608
2,048
32
65,536
complete
true
11
20
527.731079
124,032.721392
0.003846
124,184.461737
158.9694
989.5
16.065629
989.5
16.065629
81.934764
96.374005
82.397102
96.917821
FLASH_ATTENTION
null
h100_sxm_training_seed22_v2.json
baseline_reused
NVIDIA H100 80GB HBM3
GPU-8f62317d-f779-de5a-419c-ecd52f14c564
22
150m
150M
150,436,608
2,048
64
null
oom
null
null
null
null
null
null
null
null
989.5
null
989.5
null
84.017376
98.823634
84.18807
99.024408
null
CUDA out of memory. Tried to allocate 384.00 MiB. GPU 0 has a total capacity of 79.18 GiB of which 120.19 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 78.25 GiB is allocated by PyTorch, and 82.79 MiB is reserved by PyTorch but unallocated. If reserved but ...
h200_training_seed11_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
11
150m
150M
150,436,608
2,048
1
2,048
complete
true
31
20
53.839952
37,618.324291
0.01945
38,038.666568
48.693564
989.5
4.921027
989.5
4.921027
4.670883
3.111417
4.720689
3.144595
FLASH_ATTENTION
null
h200_training_seed11_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
11
150m
150M
150,436,608
2,048
2
4,096
complete
true
29
20
54.377327
75,289.598096
0.00286
75,325.512095
96.424716
989.5
9.744792
989.5
9.744792
7.163282
4.77168
7.493124
4.991398
FLASH_ATTENTION
null
h200_training_seed11_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
11
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
66.153568
123,829.782233
0.00159
123,833.078313
158.519591
989.5
16.020171
989.5
16.020171
12.148047
8.092184
12.73181
8.481045
FLASH_ATTENTION
null
h200_training_seed11_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
11
150m
150M
150,436,608
2,048
8
16,384
complete
true
11
20
114.65683
142,881.327468
0.000457
142,895.979452
182.922145
989.5
18.486321
989.5
18.486321
22.117578
14.733191
23.353885
15.556732
FLASH_ATTENTION
null
h200_training_seed11_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
11
150m
150M
150,436,608
2,048
16
32,768
complete
true
11
20
210.024414
156,011.770315
0.000318
156,019.956757
199.72224
989.5
20.184158
989.5
20.184158
42.05664
28.015206
44.459622
29.615906
FLASH_ATTENTION
null
h200_training_seed11_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
11
150m
150M
150,436,608
2,048
32
65,536
complete
true
11
20
399.628494
163,919.643697
0.001206
163,992.31021
209.927706
989.5
21.215534
989.5
21.215534
81.934764
54.579236
86.549463
57.653228
FLASH_ATTENTION
null
h200_training_seed11_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
11
150m
150M
150,436,608
2,048
64
null
oom
null
null
null
null
null
null
null
null
989.5
null
989.5
null
136.537473
90.95176
137.298444
91.458666
null
CUDA out of memory. Tried to allocate 15.62 GiB. GPU 0 has a total capacity of 139.81 GiB of which 11.80 GiB is free. Including non-PyTorch memory, this process has 128.01 GiB memory in use. Of the allocated memory 127.16 GiB is allocated by PyTorch, and 125.72 MiB is reserved by PyTorch but unallocated. If reserved bu...
h200_training_seed22_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
22
150m
150M
150,436,608
2,048
1
2,048
complete
true
44
20
55.057169
37,089.381932
0.003154
37,197.699022
47.617035
989.5
4.812232
989.5
4.812232
4.670883
3.111417
4.720689
3.144595
FLASH_ATTENTION
null
h200_training_seed22_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
22
150m
150M
150,436,608
2,048
2
4,096
complete
true
26
20
54.53648
74,907.851148
0.013372
75,105.690792
96.143321
989.5
9.716354
989.5
9.716354
7.163282
4.77168
7.493124
4.991398
FLASH_ATTENTION
null
h200_training_seed22_v2.json
baseline_reused
NVIDIA H200
GPU-b23aad41-8254-c13e-e269-0db430f83167
22
150m
150M
150,436,608
2,048
4
8,192
complete
true
11
20
65.87587
124,055.128751
0.002349
124,355.094376
159.187828
989.5
16.087704
989.5
16.087704
12.148047
8.092184
12.73181
8.481045
FLASH_ATTENTION
null