AbhishekGhosh1998 commited on
Commit
9d350fe
·
verified ·
1 Parent(s): 5f06336

granularity results granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb @ 2026-09-21T08:36:48Z from sp-aiinfra-interactive-aghosh-8-mb99k-8xb200-moe-cg-0

Browse files
Files changed (13) hide show
  1. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/manifest.txt +19 -0
  2. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/pip_freeze.txt +317 -0
  3. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/provenance.txt +42 -0
  4. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1/mem_samples.csv +216 -0
  5. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1/run.env +102 -0
  6. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1/train.log +1188 -0
  7. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2/mem_samples.csv +224 -0
  8. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2/run.env +102 -0
  9. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2/train.log +1188 -0
  10. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3/mem_samples.csv +224 -0
  11. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3/run.env +102 -0
  12. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3/train.log +1220 -0
  13. granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/results.tsv +4 -0
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/manifest.txt ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ timestamp: 2026-09-21T08:33:14+00:00
2
+ host: sp-aiinfra-interactive-aghosh-8-mb99k-8xb200-moe-cg-0
3
+ gpu: NVIDIA B200 x8
4
+ driver: 580.95.05
5
+ study_scripts_branch: granularity-echo
6
+ study_scripts_commit: b8bd100
7
+ campaign: echo
8
+ backend: echo
9
+ phase: tierb
10
+ gbs: 8
11
+ seq: 4096
12
+ configs: G1 G2 G3 G4
13
+ capacity: 1.25
14
+ repeats: 3
15
+ nsys: 1
16
+ megatron_repo: /workspace/Megatron-MoE-Echo
17
+ megatron_branch_required: bracing-regions-through-kernels
18
+ megatron_branch_actual: bracing-regions-through-kernels
19
+ megatron_commit: 9a236793
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/pip_freeze.txt ADDED
@@ -0,0 +1,317 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ absl-py==2.3.0
2
+ aiohappyeyeballs @ file:///rapids/aiohappyeyeballs-2.6.1-py3-none-any.whl#sha256=f349ba8f4b75cb25c99c5c2d84e997e485204d2902a9597802b0371f09331fb8
3
+ aiohttp @ file:///rapids/aiohttp-3.12.7-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=94f98e0e5a49f89b252e115844f756c04fc8050f38252a32a3dd994ce8121f10
4
+ aiosignal @ file:///rapids/aiosignal-1.3.2-py2.py3-none-any.whl#sha256=45cde58e409a301715980c2b01d0c28bdde3770d8290b5eb2173759d9acb31a5
5
+ annotated-types==0.7.0
6
+ anyio==4.9.0
7
+ apex @ file:///opt/pytorch/apex/dist/apex-0.1-cp312-cp312-linux_x86_64.whl#sha256=a1508e4a386362e9c8f46c92ccada9110e421212b964c694235d0fd5a224b170
8
+ argon2-cffi==25.1.0
9
+ argon2-cffi-bindings==21.2.0
10
+ arrow==1.3.0
11
+ asciitree @ file:///rapids/asciitree-0.3.3-py3-none-any.whl#sha256=3e507ccfaeb417823158091fdb9d26b0205f6a59ba680b479a71db65c73d01d4
12
+ asttokens==3.0.0
13
+ astunparse==1.6.3
14
+ async-lru==2.0.5
15
+ attrs==25.3.0
16
+ audioread==3.0.1
17
+ babel==2.17.0
18
+ beautifulsoup4==4.13.4
19
+ black==25.1.0
20
+ bleach==6.2.0
21
+ blis==0.7.11
22
+ cachetools @ file:///rapids/cachetools-6.0.0-py3-none-any.whl#sha256=82e73ba88f7b30228b5507dce1a1f878498fc669d972aef2dde4f3a3c24f103e
23
+ catalogue==2.0.10
24
+ certifi==2025.4.26
25
+ cffi==1.17.1
26
+ charset-normalizer==3.4.2
27
+ click==8.5.0
28
+ cloudpathlib==0.21.1
29
+ cloudpickle @ file:///rapids/cloudpickle-3.1.1-py3-none-any.whl#sha256=c8c5a44295039331ee9dad40ba100a9c7297b6f988e50e87ccdf3765a668350e
30
+ cmake==3.31.6
31
+ comm==0.2.2
32
+ confection==0.1.5
33
+ contourpy==1.3.2
34
+ cuda-bindings @ file:///rapids/cuda_bindings-12.9.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=ff0e28d1e34758654b9c961e1f55e4786e49aee6a4dbceaf3cc24c46c672df7e
35
+ cuda-python @ file:///rapids/cuda_python-12.9.0-py3-none-any.whl#sha256=926acba49b2c0a0374c61b7c98f337c085199cf51cdfe4d6423c4129c20547a7
36
+ cudf @ file:///rapids/cudf-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=b779a8e8c0e8577b9b560895958496da211bfac49cdc4f11a49e8d18cb82e348
37
+ cudf-polars @ file:///rapids/cudf_polars-25.4.0-py3-none-any.whl#sha256=227b6c5147e93ea86282c8f7219118f916769fa2ca0cdd931e97a148aa66b869
38
+ cugraph @ file:///rapids/cugraph-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=b837c2dc2f7b235c702e4f8df64479b8fcbb342f848516d5faf0cdc4b5f31f29
39
+ cugraph-service-client @ file:///rapids/cugraph_service_client-25.4.0-py3-none-any.whl#sha256=7e568c3e41154b9752f5173555ffcc282ac52d72343feeb79653b2e2b1232470
40
+ cugraph-service-server @ file:///rapids/cugraph_service_server-25.4.0-py3-none-any.whl#sha256=a2830dcffafcccb16e12ebf85b98f8f247869d38c01ab4dd8664a1e7439d31be
41
+ cuml @ file:///rapids/cuml-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=1e90a464a18b094e24797f1676688c5d7696c4cf97c934504214a53d5aa09aa5
42
+ cupy-cuda12x @ file:///rapids/cupy_cuda12x-13.3.0-cp312-cp312-linux_x86_64.whl#sha256=6a2dfa0aa1e8189a10d1534276c08f1d2f73b747fd28bfbe7b9f4ef66d9af73f
43
+ cuvs @ file:///rapids/cuvs-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=b12f6b9f80e053a3b38542c2947d2516e468c02cad54a13d8b1a4a61d5aa7914
44
+ cycler==0.12.1
45
+ cymem==2.0.11
46
+ Cython==3.1.1
47
+ dask @ file:///rapids/dask-2025.2.0-py3-none-any.whl#sha256=f0fdeef6ceb0a06569d456c9e704f220f7f54e80f3a6ea42ab98cea6bc642b6e
48
+ dask-cuda @ file:///rapids/dask_cuda-25.4.0-py3-none-any.whl#sha256=442d87da0a95d00ca2a262b64c76e785a32ad905c82000fa2227932f81390e19
49
+ dask-cudf @ file:///rapids/dask_cudf-25.4.0-py3-none-any.whl#sha256=5e7cea0b26daea17da6e0bb457435e24460f76ba740358a3a18e239d6015c95c
50
+ datasets==5.0.1
51
+ debugpy==1.8.14
52
+ decorator==5.2.1
53
+ -e git+https://github.com/deepseek-ai/DeepEP.git@34152ae28f80bcc3ee38d7a12cb2ad87cfd4ea72#egg=deep_ep
54
+ defusedxml==0.7.1
55
+ dill==0.4.1
56
+ distributed @ file:///rapids/distributed-2025.2.0-py3-none-any.whl#sha256=368462084be47092d0a835f6c64c9b852da64dc88302b76d5f93bd5131228c67
57
+ distributed-ucxx @ file:///rapids/distributed_ucxx-0.43.0-py3-none-any.whl#sha256=94f60d0ccb0c94ba8dbfa1eaf455874d18aaab17e2ae218c38590e9fdc2d721d
58
+ dm-tree==0.1.9
59
+ einops==0.8.1
60
+ execnet==2.1.1
61
+ executing==2.2.0
62
+ expecttest==0.3.0
63
+ fasteners @ file:///rapids/fasteners-0.19-py3-none-any.whl#sha256=758819cb5d94cdedf4e836988b74de396ceacb8e2794d21f82d131fd9ee77237
64
+ fastjsonschema==2.21.1
65
+ fastrlock @ file:///rapids/fastrlock-0.8.3-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_28_x86_64.whl#sha256=5f13ec08f1adb1aa916c384b05ecb7dbebb8df9ea81abd045f60941c6283a670
66
+ filelock==3.18.0
67
+ flash_attn @ file:///opt/pytorch/flash_attn-2.7.4.post1-cp312-cp312-linux_x86_64.whl#sha256=a10455d6c56b387785b7413fd0bbcb392b42b3d4f4b88fda5d86cc8a9ce9182a
68
+ fonttools==4.58.1
69
+ fqdn==1.5.1
70
+ frozenlist @ file:///rapids/frozenlist-1.6.0-cp312-cp312-manylinux_2_5_x86_64.manylinux1_x86_64.manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=4e1be9111cb6756868ac242b3c2bd1f09d9aea09846e4f5c23715e7afb647103
71
+ fsspec==2025.5.1
72
+ gast==0.6.0
73
+ grpcio==1.62.1
74
+ h11==0.16.0
75
+ hf-xet==1.6.0
76
+ httpcore==1.0.9
77
+ httpx==0.28.1
78
+ huggingface_hub==1.32.0
79
+ hypothesis==6.130.8
80
+ idna==3.10
81
+ importlib_metadata==8.7.0
82
+ iniconfig==2.1.0
83
+ intel-openmp==2021.4.0
84
+ ipykernel==6.29.5
85
+ ipython==9.3.0
86
+ ipython_pygments_lexers==1.1.1
87
+ isoduration==20.11.0
88
+ isort==6.0.1
89
+ jedi==0.19.2
90
+ Jinja2==3.1.6
91
+ joblib==1.5.1
92
+ json5==0.12.0
93
+ jsonpointer==3.0.0
94
+ jsonschema==4.24.0
95
+ jsonschema-specifications==2025.4.1
96
+ jupyter-events==0.12.0
97
+ jupyter-lsp==2.2.5
98
+ jupyter_client==8.6.3
99
+ jupyter_core==5.8.1
100
+ jupyter_server==2.16.0
101
+ jupyter_server_terminals==0.5.3
102
+ jupyterlab==4.4.3
103
+ jupyterlab_code_formatter==3.0.2
104
+ jupyterlab_pygments==0.3.0
105
+ jupyterlab_server==2.27.3
106
+ jupyterlab_tensorboard_pro @ file:///builder/jupyterlab_tensorboard_pro-4.0.0-py2.py3-none-any.whl#sha256=6bfcb5ced3358eb72150f4b2ab91312faef2f0ca14d845e8379d8fa52df6902e
107
+ jupytext==1.17.2
108
+ kiwisolver==1.4.8
109
+ kvikio @ file:///rapids/kvikio-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=8b40fd8ea9e4a2974c04b50b7fc26b06094ac807f9a233281d0bafbaa71b1d85
110
+ langcodes==3.5.0
111
+ language_data==1.3.0
112
+ lazy_loader==0.4
113
+ libcudf @ file:///rapids/libcudf-25.4.0-py3-none-linux_x86_64.whl#sha256=238320b4958c1737bba2a609673e2539a78216b811b5498a68e00395d9b021a4
114
+ libcugraph @ file:///rapids/libcugraph-25.4.0-py3-none-linux_x86_64.whl#sha256=29085e901b2f7569ff2653bb1355fe6fad20f2fcaa585f5649da9867d8df766d
115
+ libcuml @ file:///rapids/libcuml-25.4.0-py3-none-linux_x86_64.whl#sha256=8e2e444c72b0e995e15a48da37ff12fb410a47a2baa69ba86ccc9b1a3db91e8c
116
+ libcuvs @ file:///rapids/libcuvs-25.4.0-py3-none-linux_x86_64.whl#sha256=e2f8db61c92b769c590dd752f43439ddd62b2f6592a3c337362b6c1ef220c3b9
117
+ libkvikio @ file:///rapids/libkvikio-25.4.0-py3-none-linux_x86_64.whl#sha256=da8a5713b05fc478bf40529a3e0fb61a2d605bb5a70ca4d461f96a6005dbe8b7
118
+ libraft @ file:///rapids/libraft-25.4.0-py3-none-linux_x86_64.whl#sha256=b921dec005aa4a321dd156776a18376feb47c96a5b67eec125c8a8e19edd1e24
119
+ librmm @ file:///rapids/librmm-25.4.0-py3-none-linux_x86_64.whl#sha256=ff951811bda729d4fa1109ab906225ae1f489a6ec2eeab5d1b674cc34bd4422f
120
+ librmm-cu12 @ file:///rapids/librmm_cu12-25.4.0-py3-none-any.whl#sha256=29296ef2b0e90679a0db280e86015be12d6ab9f4e62a181aa085f62dbadd06b3
121
+ librosa==0.11.0
122
+ libucxx @ file:///rapids/libucxx-0.43.0-py3-none-linux_x86_64.whl#sha256=f15849d7ac2fb68f0fd8998838a13321ddcfb0c594f7e6cbc401a3bcc4750566
123
+ lightning-thunder @ file:///opt/pytorch/lightning-thunder
124
+ lightning-utilities==0.14.3
125
+ lintrunner==0.12.7
126
+ llvmlite==0.42.0
127
+ locket @ file:///rapids/locket-1.0.0-py2.py3-none-any.whl#sha256=b6c819a722f7b6bd955b80781788e4a66a55628b858d347536b7e81325a3a5e3
128
+ looseversion==1.3.0
129
+ marisa-trie==1.2.1
130
+ Markdown==3.8
131
+ markdown-it-py==3.0.0
132
+ MarkupSafe==3.0.2
133
+ matplotlib==3.10.3
134
+ matplotlib-inline==0.1.7
135
+ mdit-py-plugins==0.4.2
136
+ mdurl==0.1.2
137
+ -e git+ssh://git@github.com/sriram-17-17/my-Megatron-MoE-Echo.git@9a23679390d7ab1fe462bc61fa31d90d393db49c#egg=megatron_core
138
+ mistune==3.1.3
139
+ mkl==2021.1.1
140
+ mkl-devel==2021.1.1
141
+ mkl-include==2021.1.1
142
+ ml_dtypes==0.6.0
143
+ mock==5.2.0
144
+ -e git+ssh://git@github.com/sriram-17-17/MoonEP.git@d8bc6667fad7177e6e09b637d773fa67f08bae93#egg=moonep
145
+ mpmath==1.3.0
146
+ msgpack==1.1.0
147
+ multidict @ file:///rapids/multidict-6.4.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=a920f9cf2abdf6e493c519492d892c362007f113c94da4c239ae88429835bad1
148
+ multiprocess==0.70.19
149
+ murmurhash==1.0.13
150
+ mypy_extensions==1.1.0
151
+ nbclient==0.10.2
152
+ nbconvert==7.16.6
153
+ nbformat==5.10.4
154
+ nest-asyncio==1.6.0
155
+ networkx==3.5
156
+ ninja==1.11.1.4
157
+ notebook==7.4.3
158
+ notebook_shim==0.2.4
159
+ numba==0.59.1
160
+ numba-cuda @ file:///rapids/numba_cuda-0.4.0-py3-none-any.whl#sha256=5b95a5f42068771e37142944d44fe844fa87f1b2f89051461a4cb0af91bcc28a
161
+ numcodecs @ file:///rapids/numcodecs-0.13.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=da2230484e6102e5fa3cc1a5dd37ca1f92dfbd183d91662074d6f7574e3e8f53
162
+ numpy==1.26.4
163
+ nvdlfw_inspect @ git+https://github.com/NVIDIA/nvidia-dlfw-inspect.git@6b60eb0e675606fb2fbbfcb12667ecbec75eaf87
164
+ nvfuser==0.2.27a0+9bf5aca
165
+ nvidia-cuda-cccl-cu12==12.9.27
166
+ nvidia-cudnn-frontend @ file:///opt/pytorch/lightning-thunder/tmp/cudnn_frontend
167
+ nvidia-cutlass-dsl==4.4.2
168
+ nvidia-cutlass-dsl-libs-base==4.4.2
169
+ nvidia-dali-cuda120==1.50.0
170
+ nvidia-ml-py @ file:///rapids/nvidia_ml_py-12.575.51-py3-none-any.whl#sha256=eb8641800d98ce40a22f479873f34b482e214a7e80349c63be51c3919845446e
171
+ nvidia-modelopt==0.29.0
172
+ nvidia-modelopt-core==0.29.0
173
+ nvidia-nvcomp-cu12==4.2.0.14
174
+ nvidia-nvimgcodec-cu12==0.5.0.13
175
+ nvidia-nvjpeg-cu12==12.4.0.16
176
+ nvidia-nvjpeg2k-cu12==0.8.1.40
177
+ nvidia-nvshmem-cu12==3.7.2
178
+ nvidia-nvtiff-cu12==0.5.0.67
179
+ nvidia-resiliency-ext==0.4.0
180
+ nvtx @ file:///rapids/nvtx-0.2.11-cp312-cp312-linux_x86_64.whl#sha256=73521e37cba42b4db0b3cd68fc5eb814e55c500a61eeed6edcdf6746315ac53e
181
+ nx-cugraph @ file:///rapids/nx_cugraph-25.4.0-py3-none-any.whl#sha256=e625444486cc899a77bd48e7da76a2edd95c0006dd95fbf323745feae7f33111
182
+ onnx @ file:///opt/pytorch/pytorch/third_party/onnx
183
+ onnx-ir==1.0.0
184
+ onnxscript==0.7.2
185
+ opt_einsum==3.4.0
186
+ optree==0.16.0
187
+ overrides==7.7.0
188
+ packaging==23.2
189
+ pandas @ file:///rapids/pandas-2.2.3-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=fffb8ae78d8af97f849404f21411c95062db1496aeb3e56f146f0355c9989319
190
+ pandocfilters==1.5.1
191
+ parso==0.8.4
192
+ partd @ file:///rapids/partd-1.4.2-py3-none-any.whl#sha256=978e4ac767ec4ba5b86c6eaa52e5a2a3bc748a2ca839e8cc798f1cc6ce6efb0f
193
+ pathspec==0.12.1
194
+ pexpect==4.9.0
195
+ pillow==11.2.1
196
+ platformdirs==4.3.8
197
+ pluggy==1.6.0
198
+ ply @ file:///rapids/ply-3.11-py2.py3-none-any.whl#sha256=096f9b8350b65ebd2fd1346b12452efe5b9607f7482813ffca50c22722a807ce
199
+ polars @ file:///rapids/polars-1.25.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=f7fcbb4f476784384ccda48757fca4e8c2e2c5a0a3aef3717aaf56aee4e30e09
200
+ polygraphy==0.49.20
201
+ pooch==1.8.2
202
+ preshed==3.0.10
203
+ prometheus_client==0.22.1
204
+ prompt_toolkit==3.0.51
205
+ propcache @ file:///rapids/propcache-0.3.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=563f9d8c03ad645597b8d010ef4e9eab359faeb11a0a2ac9f7b4bc8c28ebef25
206
+ protobuf==4.24.4
207
+ psutil==7.0.0
208
+ ptyprocess==0.7.0
209
+ PuLP==3.2.1
210
+ pure_eval==0.2.3
211
+ pyarrow==25.0.1
212
+ pybind11==2.13.6
213
+ pybind11_global==2.13.6
214
+ pycocotools @ git+https://github.com/nvidia/cocoapi.git@08350a157b41f0755a536441b5abec89ab934c5b#subdirectory=PythonAPI
215
+ pycparser==2.22
216
+ pydantic==2.11.5
217
+ pydantic_core==2.33.2
218
+ Pygments==2.19.1
219
+ pylibcudf @ file:///rapids/pylibcudf-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=401b8bfd022d06f84e120779ca7c3d01a28b68762aaa8f7873b10d6b1975cf6a
220
+ pylibcugraph @ file:///rapids/pylibcugraph-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=061c7389a88c815b54fd5d3a3584fb87bf693e5602f5be335260d0ba575ba557
221
+ pylibcugraphops @ file:///rapids/pylibcugraphops-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=a25ef4b22166c80718d5c7e5bddb755e81a2fe340fa5becd0ac36ec3edfb208a
222
+ pylibraft @ file:///rapids/pylibraft-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=f2ecd23fe51f9c2caa87bbbe85cb32fcaff4ecfb52ceb9a9991bc50f82782c2e
223
+ pylibwholegraph @ file:///rapids/pylibwholegraph-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=b82913d1e38bda76d8c272fbd8cc9072d987209d3d47cc93a464a37c2925aca2
224
+ pynvjitlink @ file:///rapids/pynvjitlink-0.3.0-cp312-cp312-linux_x86_64.whl#sha256=066caf1314730643d3b48c9847bd65789defdd6a7c78993176eb3d8bbe3f4919
225
+ pynvml @ file:///rapids/pynvml-12.0.0-py3-none-any.whl#sha256=fdff84b62a27dbe98e08e1a647eb77342bef1aebe0878bcd15e99a83fcbecb9e
226
+ pyparsing==3.2.3
227
+ pytest==8.1.1
228
+ pytest-flakefinder==1.1.0
229
+ pytest-rerunfailures==15.1
230
+ pytest-shard==0.1.2
231
+ pytest-xdist==3.7.0
232
+ python-dateutil==2.9.0.post0
233
+ python-hostlist==2.2.1
234
+ python-json-logger==3.3.0
235
+ pytorch-triton @ file:///tmp/dist/pytorch_triton-3.3.0%2Bgit96316ce52.nvinternal-cp312-cp312-linux_x86_64.whl#sha256=6d6921772b9dfd0243c055e0d054de42a113e92cabcfdd1d3e80a5ad9c08cf7d
236
+ pytz @ file:///rapids/pytz-2023.4-py2.py3-none-any.whl#sha256=f90ef520d95e7c46951105338d918664ebfd6f1d995bd7d153127ce90efafa6a
237
+ PyYAML==6.0.2
238
+ pyzmq==26.4.0
239
+ raft-dask @ file:///rapids/raft_dask-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=36cf37ba6ce35454c4729c2cb597a360f22780eefb11d514a16a07137da5eb6c
240
+ rapids-dask-dependency @ file:///rapids/rapids_dask_dependency-25.4.0a0-py3-none-any.whl#sha256=484faca626e7760112a7d8ad3426f5bdd720b609006d8dca87c2607076a348bd
241
+ rapids-logger @ file:///rapids/rapids_logger-0.1.18-py3-none-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl#sha256=3617b2d86291ac6cb957542895f20fc9737ace29773ca90337e6fb04c73bd120
242
+ referencing==0.36.2
243
+ regex==2024.11.6
244
+ requests==2.32.3
245
+ rfc3339-validator==0.1.4
246
+ rfc3986-validator==0.1.1
247
+ rich==14.0.0
248
+ rmm @ file:///rapids/rmm-25.4.0-cp312-cp312-linux_x86_64.whl#sha256=1824bcab3836296666980a997ddcb9c4bca24dfb873012cfa0a7ecea583ca712
249
+ rpds-py==0.25.1
250
+ safetensors==0.5.3
251
+ scikit-learn==1.6.1
252
+ scipy==1.15.3
253
+ Send2Trash==1.8.3
254
+ setuptools==78.1.1
255
+ shellingham==1.5.4
256
+ six==1.16.0
257
+ smart-open==7.1.0
258
+ sniffio==1.3.1
259
+ sortedcontainers==2.4.0
260
+ soundfile==0.13.1
261
+ soupsieve==2.7
262
+ soxr==0.5.0.post1
263
+ spacy==3.7.5
264
+ spacy-legacy==3.0.12
265
+ spacy-loggers==1.0.5
266
+ srsly==2.5.1
267
+ stack-data==0.6.3
268
+ sympy==1.14.0
269
+ tabulate==0.9.0
270
+ tbb==2021.13.1
271
+ tblib @ file:///rapids/tblib-3.1.0-py3-none-any.whl#sha256=670bb4582578134b3d81a84afa1b016128b429f3d48e6cbbaecc9d15675e984e
272
+ tensorboard==2.16.2
273
+ tensorboard-data-server==0.7.2
274
+ tensorrt @ file:///workspace/TensorRT-10.11.0.33/python/tensorrt-10.11.0.33-cp312-none-linux_x86_64.whl#sha256=c2f18f72758a3d54a71901495c8fda9e7094a8bcdd2a5b6b58967fbf05733b3a
275
+ terminado==0.18.1
276
+ thinc==8.2.5
277
+ threadpoolctl==3.6.0
278
+ thriftpy2 @ file:///rapids/thriftpy2-0.5.2-cp312-cp312-linux_x86_64.whl#sha256=c0c7d939bdc5ccf7739154959938445d5fdf72f1d6e12bd563947a36f67e2147
279
+ tinycss2==1.4.0
280
+ toolz @ file:///rapids/toolz-1.0.0-py3-none-any.whl#sha256=292c8f1c4e7516bf9086f8850935c799a874039c8bcf959d47b600e4c44a6236
281
+ torch @ file:///opt/transfer/torch-2.8.0a0%2B5228986c39.nv25.6-29946474-cp312-cp312-linux_x86_64.whl#sha256=7a3c9ff5288d8102a31296b3d62491f3f47c8820a5c9aaefc7dc0fe20e1e8ad5
282
+ torch_tensorrt @ file:///opt/pytorch/torch_tensorrt/dist/torch_tensorrt-2.8.0a0-cp312-cp312-linux_x86_64.whl#sha256=26f4d7a04e7e6e017a5b917b8b98b83720bb2bb0195f48d25f52d567d286bfe1
283
+ torchao @ file:///opt/pytorch/ao
284
+ torchprofile==0.0.4
285
+ torchvision @ file:///opt/transfer/torchvision-0.22.0a0%2B95f10a4e-29946474-cp312-cp312-manylinux_2_28_x86_64.whl#sha256=746a3c674cc61680a86f4a14b374b21af1b5c21f629c42765b40c9494f6b234d
286
+ tornado==6.5.1
287
+ tqdm==4.67.1
288
+ traitlets==5.14.3
289
+ transformer_engine @ file:///pvcdata/aiinfra/b-aghosh/wheels/transformer_engine-2.10.0.dev0%2B10f6128d-cp312-cp312-linux_x86_64.whl#sha256=35cf5f4b83b2baad864e12e512bb422e0929409ac05610812fa521d4942d0189
290
+ treelite @ file:///rapids/treelite-4.4.1-py3-none-linux_x86_64.whl#sha256=174741724205540a1a5b37da81483aeff55d89548f42b3461cf07acac7ae96eb
291
+ typer==0.16.0
292
+ types-dataclasses==0.6.6
293
+ types-python-dateutil==2.9.0.20250516
294
+ typing-inspection==0.4.1
295
+ typing_extensions==4.14.0
296
+ tzdata @ file:///rapids/tzdata-2025.2-py2.py3-none-any.whl#sha256=1a403fada01ff9221ca8044d701868fa132215d84beb92242d9acd2147f667a8
297
+ ucx-py @ file:///rapids/ucx_py-0.43.0-cp312-cp312-linux_x86_64.whl#sha256=4a452fd8d8c983283ada3d49086e3657dc2ee94fbad08efdd708d15ab6e0b73a
298
+ ucxx @ file:///rapids/ucxx-0.43.0-cp312-cp312-linux_x86_64.whl#sha256=f1b4e55cb45c66fe88a2d8f9c2b87751165f3cfec21fbdc9c47c4ddff75e8a5f
299
+ -e git+https://github.com/Dots-Infra/UltraEP.git@94cab099b44fffa99a82fea99e7c12d89cf65e4f#egg=ultra_ep
300
+ uri-template==1.3.0
301
+ urllib3==2.0.7
302
+ wasabi==1.1.3
303
+ wcwidth==0.2.13
304
+ weasel==0.4.1
305
+ webcolors==24.11.1
306
+ webencodings==0.5.1
307
+ websocket-client==1.8.0
308
+ Werkzeug==3.1.3
309
+ wheel==0.45.1
310
+ wrapt==1.17.2
311
+ xdoctest==1.0.2
312
+ xgboost @ file:///rapids/xgboost-2.1.4-py3-none-manylinux_2_39_x86_64.whl#sha256=25eeca27f4048fc491940b3a0b4b5782ee48c33cb3aec059cef71f8da542930b
313
+ xxhash==4.0.1
314
+ yarl @ file:///rapids/yarl-1.20.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl#sha256=bf099e2432131093cc611623e0b0bcc399b8cddd9a91eded8bfb50402ec35018
315
+ zarr @ file:///rapids/zarr-2.18.7-py3-none-any.whl#sha256=ac3dc4033e9ae4e9d7b5e27c97ea3eaf1003cc0a07f010bd83d5134bf8c4b223
316
+ zict @ file:///rapids/zict-3.0.0-py2.py3-none-any.whl#sha256=5796e36bd0e0cc8cf0fbc1ace6a68912611c1dbd74750a3f3026b9b9d6a327ae
317
+ zipp==3.22.0
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/provenance.txt CHANGED
@@ -46,3 +46,45 @@ command : run_ladder.sh --phase tierb --gbs 8 --out /workspace/granularity
46
  ---- toolchain -----------------------------------------------
47
  nvcc : release 12.9, V12.9.86
48
  gcc : gcc (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
46
  ---- toolchain -----------------------------------------------
47
  nvcc : release 12.9, V12.9.86
48
  gcc : gcc (Ubuntu 13.3.0-6ubuntu2~24.04) 13.3.0
49
+ python : 3.12.3
50
+ torch : 2.8.0a0+5228986c39.nv25.06 cuda=12.9 nccl=2.27.3
51
+ torch arch : ['sm_75', 'sm_80', 'sm_86', 'sm_90', 'sm_100', 'sm_120', 'compute_120']
52
+ transformer_engine: 2.10.0.dev0+10f6128d [/usr/local/lib/python3.12/dist-packages/transformer_engine/__init__.py]
53
+ deep_ep : 1.2.1+34152ae [/workspace/DeepEP/deep_ep/__init__.py]
54
+ ultra_ep : 1.0.0 [/workspace/UltraEP/ultra_ep/__init__.py]
55
+ moonep : 0.0.1 [/workspace/MoonEP/moonep/__init__.py]
56
+ megatron : 0.14.0rc5 [None]
57
+ flash_attn : 2.7.4.post1 [/usr/local/lib/python3.12/dist-packages/flash_attn/__init__.py]
58
+ numpy : 1.26.4 [/usr/local/lib/python3.12/dist-packages/numpy/__init__.py]
59
+ huggingface_hub: 1.32.0 [/usr/local/lib/python3.12/dist-packages/huggingface_hub/__init__.py]
60
+
61
+ ---- repositories --------------------------------------------
62
+ Megatron-MoE-Echo bracing-regions-through-kernels 9a23679390d7
63
+ Megatron-UltraEP ultraep-token-drop-analysis eaf829e7ad21
64
+ UltraEP main 94cab099b44f
65
+ DeepEP HEAD 34152ae28f80 (+3 untracked: build output)
66
+ MoonEP megatron-ddp-symmetric-params d8bc6667fad7
67
+ blackwell_run_scripts granularity-echo b8bd10097bb5
68
+
69
+ ---- environment ---------------------------------------------
70
+ CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0
71
+ CUDA_DRIVER_VERSION=575.57.08
72
+ CUDA_HOME=/usr/local/cuda
73
+ CUDA_MODULE_LOADING=LAZY
74
+ CUDA_VERSION=12.9.1.010
75
+ HF_REMOTE_ROOT=granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16
76
+ NCCL_VERSION=2.27.3
77
+ NVSHMEM_VERSION=3.2.5
78
+ PYTORCH_BUILD_NUMBER=0
79
+ PYTORCH_BUILD_VERSION=2.8.0a0+5228986
80
+ PYTORCH_HOME=/opt/pytorch/pytorch
81
+ PYTORCH_VERSION=2.8.0a0+5228986
82
+ TORCH_ALLOW_TF32_CUBLAS_OVERRIDE=1
83
+ TORCH_CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0+PTX
84
+ TORCH_NCCL_USE_COMM_NONBLOCKING=0
85
+
86
+ ---- shared library resolution -------------------------------
87
+ libnvshmem_host.so: not in ldconfig
88
+ libnccl.so: /lib/x86_64-linux-gnu/libnccl.so.2
89
+ lrwxrwxrwx 1 root root 20 Sep 21 04:57 /usr/local/lib/python3.12/dist-packages/nvidia/nvshmem/lib/libnvshmem_host.so -> libnvshmem_host.so.3
90
+ -rw-r--r-- 1 root root 179045744 Sep 21 04:54 /usr/local/lib/python3.12/dist-packages/nvidia/nvshmem/lib/libnvshmem_host.so.3
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1/mem_samples.csv ADDED
@@ -0,0 +1,216 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 0, 0
2
+ 1, 0
3
+ 2, 0
4
+ 3, 0
5
+ 4, 0
6
+ 5, 0
7
+ 6, 0
8
+ 7, 0
9
+ 0, 0
10
+ 1, 0
11
+ 2, 0
12
+ 3, 0
13
+ 4, 0
14
+ 5, 0
15
+ 6, 0
16
+ 7, 0
17
+ 0, 4
18
+ 1, 1
19
+ 2, 1
20
+ 3, 1
21
+ 4, 1
22
+ 5, 1
23
+ 6, 1
24
+ 7, 1
25
+ 0, 4
26
+ 1, 4
27
+ 2, 4
28
+ 3, 4
29
+ 4, 4
30
+ 5, 1
31
+ 6, 1
32
+ 7, 1
33
+ 0, 4
34
+ 1, 4
35
+ 2, 4
36
+ 3, 4
37
+ 4, 4
38
+ 5, 4
39
+ 6, 4
40
+ 7, 4
41
+ 0, 27
42
+ 1, 725
43
+ 2, 725
44
+ 3, 725
45
+ 4, 725
46
+ 5, 725
47
+ 6, 725
48
+ 7, 725
49
+ 0, 907
50
+ 1, 939
51
+ 2, 917
52
+ 3, 923
53
+ 4, 929
54
+ 5, 917
55
+ 6, 937
56
+ 7, 911
57
+ 0, 1803
58
+ 1, 1803
59
+ 2, 1803
60
+ 3, 1803
61
+ 4, 1803
62
+ 5, 1803
63
+ 6, 1803
64
+ 7, 1803
65
+ 0, 13079
66
+ 1, 13105
67
+ 2, 13137
68
+ 3, 13091
69
+ 4, 13101
70
+ 5, 13111
71
+ 6, 13147
72
+ 7, 13129
73
+ 0, 13669
74
+ 1, 13671
75
+ 2, 13677
76
+ 3, 13693
77
+ 4, 13693
78
+ 5, 13669
79
+ 6, 13685
80
+ 7, 13689
81
+ 0, 14831
82
+ 1, 14833
83
+ 2, 14833
84
+ 3, 14835
85
+ 4, 14829
86
+ 5, 14833
87
+ 6, 14835
88
+ 7, 14829
89
+ 0, 14915
90
+ 1, 14917
91
+ 2, 14917
92
+ 3, 14919
93
+ 4, 14913
94
+ 5, 14917
95
+ 6, 14919
96
+ 7, 14913
97
+ 0, 15053
98
+ 1, 15055
99
+ 2, 15055
100
+ 3, 15057
101
+ 4, 15049
102
+ 5, 15099
103
+ 6, 15107
104
+ 7, 15079
105
+ 0, 15711
106
+ 1, 15713
107
+ 2, 15713
108
+ 3, 15715
109
+ 4, 15709
110
+ 5, 15713
111
+ 6, 15715
112
+ 7, 15709
113
+ 0, 15851
114
+ 1, 15853
115
+ 2, 15853
116
+ 3, 15855
117
+ 4, 15849
118
+ 5, 15853
119
+ 6, 15855
120
+ 7, 15849
121
+ 0, 16349
122
+ 1, 16351
123
+ 2, 16351
124
+ 3, 16353
125
+ 4, 16347
126
+ 5, 16351
127
+ 6, 16351
128
+ 7, 16347
129
+ 0, 16349
130
+ 1, 16351
131
+ 2, 16351
132
+ 3, 16353
133
+ 4, 16347
134
+ 5, 16351
135
+ 6, 16351
136
+ 7, 16347
137
+ 0, 28579
138
+ 1, 28581
139
+ 2, 28591
140
+ 3, 28583
141
+ 4, 28577
142
+ 5, 28581
143
+ 6, 28583
144
+ 7, 28577
145
+ 0, 28963
146
+ 1, 28965
147
+ 2, 28975
148
+ 3, 28967
149
+ 4, 28961
150
+ 5, 28965
151
+ 6, 28967
152
+ 7, 28961
153
+ 0, 28963
154
+ 1, 28965
155
+ 2, 28975
156
+ 3, 28967
157
+ 4, 28961
158
+ 5, 28965
159
+ 6, 28967
160
+ 7, 28961
161
+ 0, 28963
162
+ 1, 28965
163
+ 2, 28975
164
+ 3, 28967
165
+ 4, 28961
166
+ 5, 28965
167
+ 6, 28967
168
+ 7, 28961
169
+ 0, 28963
170
+ 1, 28965
171
+ 2, 28975
172
+ 3, 28967
173
+ 4, 28961
174
+ 5, 28965
175
+ 6, 28967
176
+ 7, 28961
177
+ 0, 29887
178
+ 1, 30399
179
+ 2, 30151
180
+ 3, 29715
181
+ 4, 29797
182
+ 5, 29923
183
+ 6, 29891
184
+ 7, 29883
185
+ 0, 30921
186
+ 1, 31429
187
+ 2, 31193
188
+ 3, 30767
189
+ 4, 30883
190
+ 5, 30917
191
+ 6, 30937
192
+ 7, 30935
193
+ 0, 42895
194
+ 1, 43043
195
+ 2, 42911
196
+ 3, 42905
197
+ 4, 43035
198
+ 5, 42903
199
+ 6, 42905
200
+ 7, 43035
201
+ 0, 42099
202
+ 1, 42311
203
+ 2, 42761
204
+ 3, 42493
205
+ 4, 42885
206
+ 5, 42491
207
+ 6, 42755
208
+ 7, 42559
209
+ 0, 563
210
+ 1, 566
211
+ 2, 566
212
+ 3, 566
213
+ 4, 566
214
+ 5, 822
215
+ 6, 42109
216
+ 7, 1846
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1/run.env ADDED
@@ -0,0 +1,102 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ backend=echo
2
+ config=OLMOE_1B_7B_L16
3
+ family=tierB
4
+ source_model=OLMoE-1B-7B-0924
5
+ standard_num_layers=16
6
+ base_rung=
7
+ repo=/workspace/Megatron-MoE-Echo
8
+ branch=bracing-regions-through-kernels
9
+ commit=9a236793
10
+ num_layers=16
11
+ hidden=2048
12
+ ffn_hidden=1024
13
+ moe_ffn_hidden=1024
14
+ num_experts=64
15
+ topk=8
16
+ experts_per_rank=8
17
+ expert_params=6291456
18
+ heads=16
19
+ kv_groups=16
20
+ kv_channels=128
21
+ qk_layernorm=0
22
+ shared_ffn=0
23
+ shared_expert_overlap=0
24
+ moe_token_dispatcher=flex
25
+ mla=0
26
+ q_lora_rank=
27
+ kv_lora_rank=512
28
+ qk_head_dim=128
29
+ qk_pos_emb_head_dim=64
30
+ v_head_dim=128
31
+ dense_first_layers=0
32
+ moe_layers=16
33
+ rotary_base=10000
34
+ rotary_percent=1.0
35
+ rotary_scaling=
36
+ mscale=
37
+ mscale_all_dim=
38
+ qkv_bias=0
39
+ router_balancing=aux_loss
40
+ aux_loss_coeff=1e-2
41
+ router_pre_softmax=1
42
+ router_fusion=0
43
+ router_dtype=fp32
44
+ router_groups=
45
+ router_group_topk=
46
+ router_score_function=softmax
47
+ router_expert_bias=0
48
+ router_bias_update_rate=
49
+ router_topk_scaling_factor=
50
+ shared_expert_gate=0
51
+ apply_layernorm_1p=0
52
+ attention_output_gate=0
53
+ linear_attention_type=
54
+ linear_attention_freq=
55
+ linear_conv_kernel_dim=
56
+ linear_key_head_dim=
57
+ linear_value_head_dim=
58
+ linear_num_key_heads=
59
+ linear_num_value_heads=
60
+ mtp_num_layers=0
61
+ mtp_loss_scaling_factor=0.1
62
+ no_weight_decay_cond_type=
63
+ max_position_embeddings=4096
64
+ init_method_std=0.02
65
+ device_initiated_grouped_gemm=1
66
+ norm_epsilon=1e-5
67
+ seq_length=4096
68
+ micro_batch_size=1
69
+ global_batch_size=8
70
+ num_microbatches=1
71
+ assignments_per_rank=32768
72
+ mean_tokens_per_expert=4096
73
+ slot_policy=matched
74
+ slots_per_rank=8
75
+ capacity_mode=static
76
+ capacity=1.08984375
77
+ cuda_graph=1
78
+ probe_mode=0
79
+ async_grad_reduce=0
80
+ branch_role=timing
81
+ memory_dump_mode=none
82
+ memory_snapshot_path=
83
+ cuda_graph_warmup_steps=3
84
+ recompute=none
85
+ force_balance=0
86
+ train_iters=12
87
+ ep_size=8
88
+ echo_dispatch_overlap=1
89
+ echo_dispatch_recompute=0
90
+ ultraep_quota_min_tokens=
91
+ ultraep_weight_sync_join=
92
+ profile_enable=0
93
+ kernel_trace=0
94
+ fp8_mode=mxfp8
95
+ fp8_recipe=mxfp8
96
+ training_seed=1234
97
+ learning_rate=1e-4
98
+ min_learning_rate=1.0e-5
99
+ lr_warmup_iters=5
100
+ state_dump_enable=0
101
+ state_dump_dir=
102
+ moonep_assert_no_pending_reduce=0
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1/train.log ADDED
@@ -0,0 +1,1188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==============================================================
2
+ OLMOE_1B_7B_L16 / echo (slots/rank=8 policy=matched)
3
+ E=64 K=8 H'=1024 m=8
4
+ L=16 seq=4096 gbs=8 (1 microbatches)
5
+ cap_mode=static cap=1.08984375 cg=1 recompute=none
6
+ memory_dump=none snapshots=/workspace/granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1/memory_snapshots/snapshot.pickle
7
+ repo=/workspace/Megatron-MoE-Echo @ 9a236793
8
+ ==============================================================
9
+ W0921 08:33:16.878000 680059 torch/distributed/run.py:766]
10
+ W0921 08:33:16.878000 680059 torch/distributed/run.py:766] *****************************************
11
+ W0921 08:33:16.878000 680059 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
+ W0921 08:33:16.878000 680059 torch/distributed/run.py:766] *****************************************
13
+ /workspace/Megatron-MoE-Echo/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-MoE-Echo/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
+ 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
18
+ WARNING: overriding default arguments for tokenizer_type:GPT2BPETokenizer with tokenizer_type:GPT2BPETokenizer
19
+ Number of virtual stages per pipeline stage: None
20
+ accumulate and all-reduce gradients in fp32 for bfloat16 data type.
21
+ using torch.bfloat16 for parameters ...
22
+ /workspace/Megatron-MoE-Echo/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
+ ------------------------ arguments ------------------------
25
+ account_for_embedding_in_pipeline_split ......... False
26
+ account_for_loss_in_pipeline_split .............. False
27
+ accumulate_allreduce_grads_in_fp32 .............. True
28
+ adam_beta1 ...................................... 0.9
29
+ adam_beta2 ...................................... 0.999
30
+ adam_eps ........................................ 1e-08
31
+ add_bias_linear ................................. False
32
+ add_position_embedding .......................... False
33
+ add_qkv_bias .................................... False
34
+ adlr_autoresume ................................. False
35
+ adlr_autoresume_interval ........................ 1000
36
+ align_grad_reduce ............................... True
37
+ align_param_gather .............................. False
38
+ app_tag_run_name ................................ None
39
+ app_tag_run_version ............................. 0.0.0
40
+ apply_layernorm_1p .............................. False
41
+ apply_query_key_layer_scaling ................... False
42
+ apply_residual_connection_post_layernorm ........ False
43
+ apply_rope_fusion ............................... True
44
+ async_save ...................................... None
45
+ async_tensor_model_parallel_allreduce ........... True
46
+ attention_backend ............................... AttnBackend.auto
47
+ attention_dropout ............................... 0.0
48
+ attention_softmax_in_fp32 ....................... False
49
+ auto_detect_ckpt_format ......................... False
50
+ barrier_with_L1_time ............................ True
51
+ bert_binary_head ................................ True
52
+ bert_embedder_type .............................. megatron
53
+ bert_load ....................................... None
54
+ bf16 ............................................ True
55
+ bias_dropout_fusion ............................. True
56
+ bias_gelu_fusion ................................ False
57
+ bias_swiglu_fusion .............................. True
58
+ biencoder_projection_dim ........................ 0
59
+ biencoder_shared_query_context_model ............ False
60
+ block_data_path ................................. None
61
+ calc_ft_timeouts ................................ False
62
+ calculate_per_token_loss ........................ False
63
+ check_for_large_grads ........................... False
64
+ check_for_nan_in_loss_and_grad .................. False
65
+ check_for_spiky_loss ............................ False
66
+ check_weight_hash_across_dp_replicas_interval ... None
67
+ ckpt_assume_constant_structure .................. False
68
+ ckpt_convert_format ............................. None
69
+ ckpt_convert_save ............................... None
70
+ ckpt_convert_update_legacy_dist_opt_format ...... False
71
+ ckpt_format ..................................... torch_dist
72
+ ckpt_fully_parallel_load ........................ False
73
+ ckpt_fully_parallel_save ........................ True
74
+ ckpt_fully_parallel_save_deprecated ............. False
75
+ ckpt_step ....................................... None
76
+ classes_fraction ................................ 1.0
77
+ clip_grad ....................................... 1.0
78
+ clone_scatter_output_in_embedding ............... True
79
+ config_logger_dir ...............................
80
+ consumed_train_samples .......................... 0
81
+ consumed_valid_samples .......................... 0
82
+ context_parallel_size ........................... 1
83
+ cp_comm_type .................................... ['p2p']
84
+ create_attention_mask_in_dataloader ............. True
85
+ cross_entropy_fusion_impl ....................... native
86
+ cross_entropy_loss_fusion ....................... False
87
+ cuda_graph_scope ................................ full_iteration
88
+ cuda_graph_warmup_steps ......................... 3
89
+ data_args_path .................................. None
90
+ data_cache_path ................................. None
91
+ data_parallel_random_init ....................... False
92
+ data_parallel_sharding_strategy ................. no_shard
93
+ data_parallel_size .............................. 8
94
+ data_path ....................................... ['/workspace/data/wikitext/wikitext103_gpt2_text_document']
95
+ data_per_class_fraction ......................... 1.0
96
+ data_sharding ................................... True
97
+ dataloader_type ................................. single
98
+ ddp_average_in_collective ....................... False
99
+ ddp_bucket_size ................................. None
100
+ ddp_num_buckets ................................. None
101
+ ddp_pad_buckets_for_high_nccl_busbw ............. False
102
+ decoder_first_pipeline_num_layers ............... None
103
+ decoder_last_pipeline_num_layers ................ None
104
+ decoder_num_layers .............................. None
105
+ decoder_seq_length .............................. None
106
+ decoupled_lr .................................... None
107
+ decoupled_min_lr ................................ None
108
+ decrease_batch_size_if_needed ................... False
109
+ defer_embedding_wgrad_compute ................... False
110
+ delay_wgrad_compute ............................. False
111
+ deprecated_use_mcore_models ..................... True
112
+ deterministic_mode .............................. False
113
+ dino_bottleneck_size ............................ 256
114
+ dino_freeze_last_layer .......................... 1
115
+ dino_head_hidden_size ........................... 2048
116
+ dino_local_crops_number ......................... 10
117
+ dino_local_img_size ............................. 96
118
+ dino_norm_last_layer ............................ False
119
+ dino_teacher_temp ............................... 0.07
120
+ dino_warmup_teacher_temp ........................ 0.04
121
+ dino_warmup_teacher_temp_epochs ................. 30
122
+ disable_bf16_reduced_precision_matmul ........... False
123
+ disable_mamba_mem_eff_path ...................... False
124
+ disable_straggler_on_startup .................... False
125
+ dist_ckpt_format_deprecated ..................... None
126
+ dist_ckpt_strictness ............................ assume_ok_unexpected
127
+ distribute_saved_activations .................... False
128
+ distributed_backend ............................. nccl
129
+ distributed_timeout_minutes ..................... 10
130
+ embedding_init_method_std ....................... None
131
+ embedding_path .................................. None
132
+ empty_unused_memory_level ....................... 0
133
+ enable_cuda_graph ............................... True
134
+ enable_experimental ............................. False
135
+ enable_ft_package ............................... False
136
+ enable_gloo_process_groups ...................... True
137
+ enable_msc ...................................... True
138
+ enable_one_logger ............................... True
139
+ encoder_num_layers .............................. 16
140
+ encoder_seq_length .............................. 4096
141
+ end_weight_decay ................................ 0.1
142
+ eod_mask_loss ................................... False
143
+ error_injection_rate ............................ 0
144
+ error_injection_type ............................ transient_error
145
+ eval_interval ................................... 1000000
146
+ eval_iters ...................................... 0
147
+ evidence_data_path .............................. None
148
+ exit_duration_in_mins ........................... None
149
+ exit_interval ................................... None
150
+ exit_on_missing_checkpoint ...................... False
151
+ exit_signal_handler ............................. False
152
+ exp_avg_dtype ................................... torch.float32
153
+ exp_avg_sq_dtype ................................ torch.float32
154
+ expert_model_parallel_size ...................... 8
155
+ expert_tensor_parallel_size ..................... 1
156
+ export_force_local_attention .................... False
157
+ export_kd_cfg ................................... None
158
+ export_kd_teacher_ckpt_format ................... None
159
+ export_kd_teacher_load .......................... None
160
+ export_kv_cache_quant ........................... False
161
+ export_legacy_megatron .......................... False
162
+ export_model_type ............................... GPTModel
163
+ export_moe_apply_probs_on_input ................. False
164
+ export_qk_l2_norm ............................... False
165
+ export_quant_cfg ................................ None
166
+ export_real_quant_cfg ........................... None
167
+ export_te_mcore_model ........................... False
168
+ external_cuda_graph ............................. False
169
+ ffn_hidden_size ................................. 1024
170
+ finetune ........................................ False
171
+ finetune_data_split ............................. train
172
+ finetune_hf_dataset ............................. None
173
+ first_last_layers_bf16 .......................... False
174
+ flash_decode .................................... False
175
+ fp16 ............................................ False
176
+ fp16_lm_cross_entropy ........................... False
177
+ fp32_residual_connection ........................ False
178
+ fp8 ............................................. e4m3
179
+ fp8_amax_compute_algo ........................... most_recent
180
+ fp8_amax_history_len ............................ 1
181
+ fp8_interval .................................... 1
182
+ fp8_margin ...................................... 0
183
+ fp8_param_gather ................................ False
184
+ fp8_recipe ...................................... mxfp8
185
+ fp8_wgrad ....................................... True
186
+ fsdp_double_buffer .............................. False
187
+ global_batch_size ............................... 8
188
+ grad_reduce_in_bf16 ............................. False
189
+ gradient_accumulation_fusion .................... True
190
+ gradient_reduce_div_fusion ...................... True
191
+ group_query_attention ........................... True
192
+ head_lr_mult .................................... 1.0
193
+ heterogeneous_layers_config_encoded_json ........ None
194
+ heterogeneous_layers_config_path ................ None
195
+ hidden_dropout .................................. 0.0
196
+ hidden_size ..................................... 2048
197
+ hierarchical_context_parallel_sizes ............. None
198
+ high_priority_stream_groups ..................... []
199
+ hybrid_attention_ratio .......................... 0.0
200
+ hybrid_mlp_ratio ................................ 0.0
201
+ hybrid_override_pattern ......................... None
202
+ hysteresis ...................................... 2
203
+ ict_head_size ................................... None
204
+ ict_load ........................................ None
205
+ img_h ........................................... 224
206
+ img_w ........................................... 224
207
+ indexer_batch_size .............................. 128
208
+ indexer_log_interval ............................ 1000
209
+ inference_batch_times_seqlen_threshold .......... -1
210
+ inference_dynamic_batching ...................... False
211
+ inference_dynamic_batching_buffer_guaranteed_fraction 0.2
212
+ inference_dynamic_batching_buffer_overflow_factor None
213
+ inference_dynamic_batching_buffer_size_gb ....... 40.0
214
+ inference_dynamic_batching_chunk_size ........... 256
215
+ inference_dynamic_batching_max_requests_override None
216
+ inference_dynamic_batching_max_tokens_override .. None
217
+ inference_dynamic_batching_num_cuda_graphs ...... 16
218
+ inference_max_batch_size ........................ 8
219
+ inference_max_seq_length ........................ 2560
220
+ inference_rng_tracker ........................... False
221
+ init_method_std ................................. 0.02
222
+ init_method_xavier_uniform ...................... False
223
+ init_model_with_meta_device ..................... False
224
+ initial_loss_scale .............................. 4294967296
225
+ inprocess_active_world_size ..................... 8
226
+ inprocess_barrier_timeout ....................... 120
227
+ inprocess_completion_timeout .................... 120
228
+ inprocess_empty_cuda_cache ...................... False
229
+ inprocess_granularity ........................... node
230
+ inprocess_hard_timeout .......................... 90
231
+ inprocess_heartbeat_interval .................... 30
232
+ inprocess_heartbeat_timeout ..................... 60
233
+ inprocess_last_call_wait ........................ 1
234
+ inprocess_max_iterations ........................ None
235
+ inprocess_monitor_process_interval .............. 1.0
236
+ inprocess_monitor_thread_interval ............... 1.0
237
+ inprocess_progress_watchdog_interval ............ 1.0
238
+ inprocess_restart ............................... False
239
+ inprocess_soft_timeout .......................... 60
240
+ inprocess_termination_grace_time ................ 1
241
+ is_hybrid_model ................................. False
242
+ iter_per_epoch .................................. 1250
243
+ iterations_to_skip .............................. []
244
+ keep_fp8_transpose_cache_when_using_custom_fsdp . False
245
+ kitchen_config_file ............................. None
246
+ kitchen_recipe_number ........................... None
247
+ kv_channels ..................................... 128
248
+ kv_lora_rank .................................... 32
249
+ lazy_mpu_init ................................... None
250
+ load ............................................ None
251
+ load_main_params_from_ckpt ...................... None
252
+ load_model_opt_format ........................... False
253
+ local_rank ...................................... 0
254
+ log_energy ...................................... False
255
+ log_interval .................................... 1
256
+ log_loss_scale_to_tensorboard ................... True
257
+ log_memory_to_tensorboard ....................... False
258
+ log_num_zeros_in_grad ........................... False
259
+ log_params_norm ................................. False
260
+ log_progress .................................... False
261
+ log_straggler ................................... False
262
+ log_throughput .................................. True
263
+ log_timers_to_tensorboard ....................... False
264
+ log_validation_ppl_to_tensorboard ............... False
265
+ log_world_size_to_tensorboard ................... False
266
+ logging_level ................................... None
267
+ loss_scale ...................................... None
268
+ loss_scale_window ............................... 1000
269
+ lr .............................................. 0.0001
270
+ lr_decay_iters .................................. 8
271
+ lr_decay_samples ................................ None
272
+ lr_decay_style .................................. cosine
273
+ lr_warmup_fraction .............................. None
274
+ lr_warmup_init .................................. 0.0
275
+ lr_warmup_iters ................................. 5
276
+ lr_warmup_samples ............................... 0
277
+ lr_wsd_decay_iters .............................. None
278
+ lr_wsd_decay_samples ............................ None
279
+ lr_wsd_decay_style .............................. exponential
280
+ main_grads_dtype ................................ torch.float32
281
+ main_params_dtype ............................... torch.float32
282
+ make_vocab_size_divisible_by .................... 128
283
+ mamba_head_dim .................................. 64
284
+ mamba_num_groups ................................ 8
285
+ mamba_num_heads ................................. None
286
+ mamba_state_dim ................................. 128
287
+ manual_gc ....................................... False
288
+ manual_gc_eval .................................. True
289
+ manual_gc_interval .............................. 0
290
+ mask_factor ..................................... 1.0
291
+ mask_prob ....................................... 0.15
292
+ mask_type ....................................... random
293
+ masked_softmax_fusion ........................... False
294
+ max_position_embeddings ......................... 4096
295
+ max_tokens_to_oom ............................... 12000
296
+ memory_snapshot_path ............................ snapshot.pickle
297
+ merge_file ...................................... /workspace/data/wikitext/gpt2-merges.txt
298
+ micro_batch_size ................................ 1
299
+ microbatch_group_size_per_vp_stage .............. None
300
+ mid_level_dataset_surplus ....................... 0.005
301
+ min_loss_scale .................................. 1.0
302
+ min_lr .......................................... 1e-05
303
+ mlp_chunks_for_prefill .......................... 1
304
+ mmap_bin_files .................................. True
305
+ mock_data ....................................... False
306
+ moe_apply_probs_on_input ........................ False
307
+ moe_aux_loss_coeff .............................. 0.01
308
+ moe_deepep_num_sms .............................. 20
309
+ moe_echo_enable_random_offloading ............... False
310
+ moe_echo_expert_dispatch_overlap ................ True
311
+ moe_echo_expert_dispatcher_type ................. hybridep
312
+ moe_echo_recompute_expert_dispatch .............. False
313
+ moe_enable_deepep ............................... False
314
+ moe_enable_echo ................................. True
315
+ moe_enable_hybridep ............................. True
316
+ moe_expert_capacity_factor ...................... None
317
+ moe_extended_tp ................................. False
318
+ moe_ffn_hidden_size ............................. 1024
319
+ moe_grouped_gemm ................................ True
320
+ moe_hybridep_num_sms ............................ 16
321
+ moe_input_jitter_eps ............................ None
322
+ moe_layer_freq .................................. 1
323
+ moe_layer_recompute ............................. False
324
+ moe_num_echo_experts ............................ 64
325
+ moe_pad_expert_input_to_capacity ................ False
326
+ moe_per_layer_logging ........................... False
327
+ moe_permute_fusion .............................. False
328
+ moe_received_token_capacity ..................... 1.08984375
329
+ moe_router_bias_update_rate ..................... 0.001
330
+ moe_router_dtype ................................ fp32
331
+ moe_router_enable_expert_bias ................... False
332
+ moe_router_force_load_balancing ................. False
333
+ moe_router_group_topk ........................... None
334
+ moe_router_load_balancing_type .................. aux_loss
335
+ moe_router_num_groups ........................... None
336
+ moe_router_padding_for_fp8 ...................... False
337
+ moe_router_pre_softmax .......................... True
338
+ moe_router_score_function ....................... softmax
339
+ moe_router_topk ................................. 8
340
+ moe_router_topk_scaling_factor .................. None
341
+ moe_shared_expert_intermediate_size ............. None
342
+ moe_shared_expert_overlap ....................... False
343
+ moe_token_dispatcher_type ....................... flex
344
+ moe_token_drop_policy ........................... probs
345
+ moe_upcycling_granularity ....................... 1
346
+ moe_use_device_initiated_grouped_gemm ........... True
347
+ moe_use_legacy_grouped_gemm ..................... False
348
+ moe_use_upcycling ............................... False
349
+ moe_z_loss_coeff ................................ None
350
+ mrope_section ................................... None
351
+ mscale .......................................... 1.0
352
+ mscale_all_dim .................................. 0.0
353
+ mtp_loss_scaling_factor ......................... 0.1
354
+ mtp_num_layers .................................. None
355
+ multi_latent_attention .......................... False
356
+ nccl_all_reduce_for_prefill ..................... False
357
+ nccl_communicator_config_path ................... None
358
+ nccl_ub ......................................... False
359
+ no_load_optim ................................... None
360
+ no_load_rng ..................................... None
361
+ no_persist_layer_norm ........................... False
362
+ no_rope_freq .................................... None
363
+ no_save_optim ................................... None
364
+ no_save_rng ..................................... None
365
+ non_persistent_ckpt_type ........................ None
366
+ non_persistent_global_ckpt_dir .................. None
367
+ non_persistent_local_ckpt_algo .................. fully_parallel
368
+ non_persistent_local_ckpt_dir ................... None
369
+ non_persistent_save_interval .................... None
370
+ norm_epsilon .................................... 1e-05
371
+ normalization ................................... RMSNorm
372
+ num_attention_heads ............................. 16
373
+ num_channels .................................... 3
374
+ num_classes ..................................... 1000
375
+ num_dataset_builder_threads ..................... 1
376
+ num_distributed_optimizer_instances ............. 1
377
+ num_experts ..................................... 64
378
+ num_layers ...................................... 16
379
+ num_layers_at_end_in_bf16 ....................... 1
380
+ num_layers_at_start_in_bf16 ..................... 1
381
+ num_layers_per_virtual_pipeline_stage ........... None
382
+ num_query_groups ................................ 16
383
+ num_virtual_stages_per_pipeline_rank ............ None
384
+ num_workers ..................................... 2
385
+ object_storage_cache_path ....................... None
386
+ one_logger_async ................................ False
387
+ one_logger_project .............................. megatron-lm
388
+ one_logger_run_name ............................. None
389
+ onnx_safe ....................................... None
390
+ openai_gelu ..................................... False
391
+ optimizer ....................................... adam
392
+ optimizer_cpu_offload ........................... False
393
+ optimizer_offload_fraction ...................... 1.0
394
+ output_bert_embeddings .......................... False
395
+ overlap_cpu_optimizer_d2h_h2d ................... False
396
+ overlap_grad_reduce ............................. True
397
+ overlap_moe_expert_parallel_comm ................ False
398
+ overlap_p2p_comm ................................ False
399
+ overlap_p2p_comm_warmup_flush ................... False
400
+ overlap_param_gather ............................ True
401
+ overlap_param_gather_with_optimizer_step ........ False
402
+ override_opt_param_scheduler .................... False
403
+ params_dtype .................................... torch.bfloat16
404
+ patch_dim ....................................... 16
405
+ per_split_data_args_path ........................ None
406
+ perform_initialization .......................... True
407
+ pin_cpu_grads ................................... True
408
+ pin_cpu_params .................................. True
409
+ pipeline_model_parallel_comm_backend ............ None
410
+ pipeline_model_parallel_layout .................. None
411
+ pipeline_model_parallel_size .................... 1
412
+ position_embedding_type ......................... rope
413
+ pretrained_checkpoint ........................... None
414
+ profile ......................................... False
415
+ profile_ranks ................................... [0]
416
+ profile_step_end ................................ 12
417
+ profile_step_start .............................. 10
418
+ q_lora_rank ..................................... None
419
+ qk_head_dim ..................................... 128
420
+ qk_l2_norm ...................................... False
421
+ qk_layernorm .................................... False
422
+ qk_pos_emb_head_dim ............................. 64
423
+ query_in_block_prob ............................. 0.1
424
+ rampup_batch_size ............................... None
425
+ rank ............................................ 0
426
+ recompute_granularity ........................... None
427
+ recompute_method ................................ None
428
+ recompute_modules ............................... None
429
+ recompute_num_layers ............................ None
430
+ record_memory_history ........................... False
431
+ relative_attention_max_distance ................. 128
432
+ relative_attention_num_buckets .................. 32
433
+ replication ..................................... False
434
+ replication_factor .............................. 2
435
+ replication_jump ................................ None
436
+ rerun_mode ...................................... validate_results
437
+ reset_attention_mask ............................ False
438
+ reset_position_ids .............................. False
439
+ result_rejected_tracker_filename ................ None
440
+ retriever_report_topk_accuracies ................ []
441
+ retriever_score_scaling ......................... False
442
+ retriever_seq_length ............................ 256
443
+ retro_add_retriever ............................. False
444
+ retro_attention_gate ............................ 1
445
+ retro_cyclic_train_iters ........................ None
446
+ retro_encoder_attention_dropout ................. 0.1
447
+ retro_encoder_hidden_dropout .................... 0.1
448
+ retro_encoder_layers ............................ 2
449
+ retro_num_neighbors ............................. 2
450
+ retro_num_retrieved_chunks ...................... 2
451
+ retro_project_dir ............................... None
452
+ retro_verify_neighbor_count ..................... True
453
+ reuse_grad_buf_for_mxfp8_param_ag ............... False
454
+ rope_scaling_factor ............................. 8.0
455
+ rope_type ....................................... None
456
+ rotary_base ..................................... 10000
457
+ rotary_interleaved .............................. False
458
+ rotary_percent .................................. 1.0
459
+ rotary_scaling_factor ........................... 1.0
460
+ rotary_seq_len_interpolation_factor ............. None
461
+ run_workload_inspector_server ................... False
462
+ sample_rate ..................................... 1.0
463
+ save ............................................ None
464
+ save_interval ................................... None
465
+ save_retain_interval ............................ None
466
+ scatter_gather_tensors_in_pipeline .............. True
467
+ seed ............................................ 1234
468
+ seq_length ...................................... 4096
469
+ sequence_parallel ............................... False
470
+ sft ............................................. False
471
+ sft_tokenizer_prompt_format ..................... nemotron-h-aligned
472
+ sgd_momentum .................................... 0.9
473
+ short_seq_prob .................................. 0.1
474
+ skip_train ...................................... False
475
+ skipped_train_samples ........................... 0
476
+ spec ............................................ None
477
+ split ........................................... 100,0,0
478
+ squared_relu .................................... False
479
+ start_weight_decay .............................. 0.1
480
+ straggler_ctrlr_port ............................ 65535
481
+ straggler_minmax_count .......................... 1
482
+ suggested_communication_unit_size ............... None
483
+ swiglu .......................................... True
484
+ swin_backbone_type .............................. tiny
485
+ symmetric_ar_type ............................... None
486
+ te_rng_tracker .................................. True
487
+ tensor_model_parallel_size ...................... 1
488
+ tensorboard_dir ................................. None
489
+ tensorboard_log_interval ........................ 1
490
+ tensorboard_queue_size .......................... 1000
491
+ test_data_path .................................. None
492
+ test_mode ....................................... False
493
+ tiktoken_num_special_tokens ..................... 1000
494
+ tiktoken_pattern ................................ None
495
+ tiktoken_special_tokens ......................... None
496
+ timing_log_level ................................ 0
497
+ timing_log_option ............................... minmax
498
+ titles_data_path ................................ None
499
+ tokenizer_model ................................. None
500
+ tokenizer_type .................................. GPT2BPETokenizer
501
+ torch_fsdp2_reshard_after_forward ............... True
502
+ tp_comm_bootstrap_backend ....................... nccl
503
+ tp_comm_bulk_dgrad .............................. True
504
+ tp_comm_bulk_wgrad .............................. True
505
+ tp_comm_overlap ................................. False
506
+ tp_comm_overlap_ag .............................. True
507
+ tp_comm_overlap_cfg ............................. None
508
+ tp_comm_overlap_rs .............................. True
509
+ tp_comm_overlap_rs_dgrad ........................ False
510
+ tp_comm_split_ag ................................ True
511
+ tp_comm_split_rs ................................ True
512
+ train_data_path ................................. None
513
+ train_iters ..................................... 12
514
+ train_samples ................................... None
515
+ train_sync_interval ............................. None
516
+ transformer_impl ................................ transformer_engine
517
+ transformer_pipeline_model_parallel_size ........ 1
518
+ untie_embeddings_and_output_weights ............. True
519
+ use_checkpoint_args ............................. False
520
+ use_checkpoint_opt_param_scheduler .............. False
521
+ use_cpu_initialization .......................... None
522
+ use_custom_fsdp ................................. False
523
+ use_dist_ckpt ................................... True
524
+ use_dist_ckpt_deprecated ........................ False
525
+ use_distributed_optimizer ....................... True
526
+ use_flash_attn .................................. False
527
+ use_legacy_models ............................... False
528
+ use_mp_args_from_checkpoint_args ................ False
529
+ use_one_sent_docs ............................... False
530
+ use_persistent_ckpt_worker ...................... False
531
+ use_precision_aware_optimizer ................... False
532
+ use_pytorch_profiler ............................ False
533
+ use_ring_exchange_p2p ........................... False
534
+ use_rope_scaling ................................ False
535
+ use_rotary_position_embeddings .................. False
536
+ use_sharp ....................................... False
537
+ use_tokenizer_model_from_checkpoint_args ........ True
538
+ use_torch_fsdp2 ................................. False
539
+ use_torch_optimizer_for_cpu_offload ............. False
540
+ use_tp_pp_dp_mapping ............................ False
541
+ v_head_dim ...................................... 128
542
+ valid_data_path ................................. None
543
+ variable_seq_lengths ............................ False
544
+ virtual_pipeline_model_parallel_size ............ None
545
+ vision_backbone_type ............................ vit
546
+ vision_pretraining .............................. False
547
+ vision_pretraining_type ......................... classify
548
+ vocab_extra_ids ................................. 0
549
+ vocab_file ...................................... /workspace/data/wikitext/gpt2-vocab.json
550
+ vocab_size ...................................... None
551
+ wandb_exp_name ..................................
552
+ wandb_project ...................................
553
+ wandb_save_dir ..................................
554
+ weight_decay .................................... 0.1
555
+ weight_decay_incr_style ......................... constant
556
+ wgrad_deferral_limit ............................ 0
557
+ world_size ...................................... 8
558
+ yaml_cfg ........................................ None
559
+ -------------------- end of arguments ---------------------
560
+ INFO:megatron.core.num_microbatches_calculator:setting number of microbatches to constant 1
561
+ > building GPT2BPETokenizer tokenizer ...
562
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
563
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
564
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
565
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
566
+ > padded vocab (size: 50257) with 47 dummy tokens (new size: 50304)
567
+ WARNING:megatron.core.rerun_state_machine:RerunStateMachine initialized in mode RerunMode.VALIDATE_RESULTS
568
+ > initializing torch distributed ...
569
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
570
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
571
+ 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
572
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
573
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
574
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
575
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
576
+ [Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
577
+ [Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
578
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
579
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
580
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7[Gloo] Rank
581
+ 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
582
+ [Gloo] Rank [Gloo] Rank 5 is connected to 7 peer ranks. 6Expected number of connected peer ranks is : is connected to 77
583
+ peer ranks. Expected number of connected peer ranks is : 7
584
+ [Gloo] Rank [Gloo] Rank 4[Gloo] Rank is connected to 307 is connected to is connected to peer ranks. 77Expected number of connected peer ranks is : peer ranks. peer ranks. 7Expected number of connected peer ranks is : Expected number of connected peer ranks is :
585
+ 77
586
+
587
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
588
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7[Gloo] Rank
589
+ 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
590
+ [Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
591
+ [Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
592
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
593
+ > initialized tensor model parallel with size 1
594
+ > initialized pipeline model parallel with size 1
595
+ > setting random seeds to 1234 ...
596
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
597
+ [Gloo] Rank [Gloo] Rank 0[Gloo] Rank is connected to 000 is connected to is connected to peer ranks. 00Expected number of connected peer ranks is : peer ranks. peer ranks. 0Expected number of connected peer ranks is : Expected number of connected peer ranks is :
598
+ 00
599
+
600
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
601
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
602
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
603
+ [rank4]:[W921 08:33:33.235041645 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.
604
+ [rank1]:[W921 08:33:33.235050303 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.
605
+ [rank2]:[W921 08:33:33.235073360 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.
606
+ [rank3]:[W921 08:33:33.235075589 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.
607
+ [rank6]:[W921 08:33:33.235143561 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.
608
+ [rank5]:[W921 08:33:33.235153918 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.
609
+ [rank7]:[W921 08:33:33.239108864 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.
610
+ > compiling dataset index builder ...
611
+ make: Entering directory '/workspace/Megatron-MoE-Echo/megatron/core/datasets'
612
+ make: Nothing to be done for 'default'.
613
+ make: Leaving directory '/workspace/Megatron-MoE-Echo/megatron/core/datasets'
614
+ >>> done with dataset index builder. Compilation time: 0.034 seconds
615
+ WARNING: constraints for invoking optimized fused softmax kernel are not met. We default back to unfused kernel invocations.
616
+ > compiling and loading fused kernels ...
617
+ [rank0]:[W921 08:33:33.409859328 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.
618
+ >>> done with compiling and loading fused kernels. Compilation time: 3.955 seconds
619
+ time to initialize megatron (seconds): 7.056
620
+ [after megatron is initialized] datetime: 2026-09-21 08:33:39
621
+ building GPT model ...
622
+ /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.
623
+ warnings.warn(
624
+ /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.
625
+ warnings.warn(
626
+ /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.
627
+ warnings.warn(
628
+ /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.
629
+ warnings.warn(
630
+ /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.
631
+ warnings.warn(
632
+ /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.
633
+ warnings.warn(
634
+ /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.
635
+ warnings.warn(
636
+ /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.
637
+ warnings.warn(
638
+ > number of parameters on (tensor, pipeline) model parallel rank (0, 0): 1281951744
639
+ 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=False, 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)
640
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 7
641
+ Params for bucket 1 (103022592 elements, 103022592 padded size):
642
+ module.output_layer.weight
643
+ Params for bucket 2 (46542848 elements, 46542848 padded size):
644
+ module.decoder.layers.15.mlp.router.weight
645
+ module.decoder.layers.15.pre_mlp_layernorm.weight
646
+ module.decoder.layers.14.mlp.router.weight
647
+ module.decoder.layers.14.self_attention.linear_qkv.weight
648
+ module.decoder.layers.14.self_attention.linear_qkv.layer_norm_weight
649
+ module.decoder.layers.14.self_attention.linear_proj.weight
650
+ module.decoder.layers.13.mlp.router.weight
651
+ module.decoder.layers.13.pre_mlp_layernorm.weight
652
+ module.decoder.final_layernorm.weight
653
+ module.decoder.layers.15.self_attention.linear_qkv.layer_norm_weight
654
+ module.decoder.layers.15.self_attention.linear_qkv.weight
655
+ module.decoder.layers.15.self_attention.linear_proj.weight
656
+ module.decoder.layers.14.pre_mlp_layernorm.weight
657
+ module.decoder.layers.13.self_attention.linear_qkv.weight
658
+ Params for bucket 3 (50737152 elements, 50737152 padded size):
659
+ module.decoder.layers.12.self_attention.linear_qkv.weight
660
+ module.decoder.layers.11.mlp.router.weight
661
+ module.decoder.layers.11.pre_mlp_layernorm.weight
662
+ module.decoder.layers.11.self_attention.linear_qkv.weight
663
+ module.decoder.layers.11.self_attention.linear_proj.weight
664
+ module.decoder.layers.10.mlp.router.weight
665
+ module.decoder.layers.10.pre_mlp_layernorm.weight
666
+ module.decoder.layers.13.self_attention.linear_qkv.layer_norm_weight
667
+ module.decoder.layers.12.mlp.router.weight
668
+ module.decoder.layers.12.pre_mlp_layernorm.weight
669
+ module.decoder.layers.13.self_attention.linear_proj.weight
670
+ module.decoder.layers.12.self_attention.linear_qkv.layer_norm_weight
671
+ module.decoder.layers.12.self_attention.linear_proj.weight
672
+ module.decoder.layers.11.self_attention.linear_qkv.layer_norm_weight
673
+ module.decoder.layers.10.self_attention.linear_qkv.weight
674
+ Params for bucket 4 (50737152 elements, 50737152 padded size):
675
+ module.decoder.layers.10.self_attention.linear_proj.weight
676
+ module.decoder.layers.9.pre_mlp_layernorm.weight
677
+ module.decoder.layers.9.self_attention.linear_qkv.weight
678
+ module.decoder.layers.9.self_attention.linear_qkv.layer_norm_weight
679
+ module.decoder.layers.8.mlp.router.weight
680
+ module.decoder.layers.8.pre_mlp_layernorm.weight
681
+ module.decoder.layers.8.self_attention.linear_qkv.weight
682
+ module.decoder.layers.8.self_attention.linear_proj.weight
683
+ module.decoder.layers.7.mlp.router.weight
684
+ module.decoder.layers.7.pre_mlp_layernorm.weight
685
+ module.decoder.layers.10.self_attention.linear_qkv.layer_norm_weight
686
+ module.decoder.layers.9.mlp.router.weight
687
+ module.decoder.layers.9.self_attention.linear_proj.weight
688
+ module.decoder.layers.8.self_attention.linear_qkv.layer_norm_weight
689
+ module.decoder.layers.7.self_attention.linear_qkv.weight
690
+ Params for bucket 5 (50737152 elements, 50737152 padded size):
691
+ module.decoder.layers.6.pre_mlp_layernorm.weight
692
+ module.decoder.layers.6.mlp.router.weight
693
+ module.decoder.layers.5.mlp.router.weight
694
+ module.decoder.layers.5.pre_mlp_layernorm.weight
695
+ module.decoder.layers.4.self_attention.linear_qkv.weight
696
+ module.decoder.layers.7.self_attention.linear_qkv.layer_norm_weight
697
+ module.decoder.layers.7.self_attention.linear_proj.weight
698
+ module.decoder.layers.6.self_attention.linear_qkv.weight
699
+ module.decoder.layers.6.self_attention.linear_qkv.layer_norm_weight
700
+ module.decoder.layers.6.self_attention.linear_proj.weight
701
+ module.decoder.layers.5.self_attention.linear_qkv.weight
702
+ module.decoder.layers.5.self_attention.linear_qkv.layer_norm_weight
703
+ module.decoder.layers.5.self_attention.linear_proj.weight
704
+ module.decoder.layers.4.mlp.router.weight
705
+ module.decoder.layers.4.pre_mlp_layernorm.weight
706
+ Params for bucket 6 (50737152 elements, 50737152 padded size):
707
+ module.decoder.layers.3.pre_mlp_layernorm.weight
708
+ module.decoder.layers.3.mlp.router.weight
709
+ module.decoder.layers.2.mlp.router.weight
710
+ module.decoder.layers.2.pre_mlp_layernorm.weight
711
+ module.decoder.layers.2.self_attention.linear_qkv.weight
712
+ module.decoder.layers.1.pre_mlp_layernorm.weight
713
+ module.decoder.layers.1.self_attention.linear_qkv.weight
714
+ module.decoder.layers.4.self_attention.linear_qkv.layer_norm_weight
715
+ module.decoder.layers.3.self_attention.linear_qkv.weight
716
+ module.decoder.layers.4.self_attention.linear_proj.weight
717
+ module.decoder.layers.3.self_attention.linear_qkv.layer_norm_weight
718
+ module.decoder.layers.3.self_attention.linear_proj.weight
719
+ module.decoder.layers.2.self_attention.linear_qkv.layer_norm_weight
720
+ module.decoder.layers.2.self_attention.linear_proj.weight
721
+ module.decoder.layers.1.mlp.router.weight
722
+ Params for bucket 7 (124131328 elements, 124131328 padded size):
723
+ module.decoder.layers.1.self_attention.linear_qkv.layer_norm_weight
724
+ module.decoder.layers.0.self_attention.linear_qkv.weight
725
+ module.decoder.layers.0.mlp.router.weight
726
+ module.embedding.word_embeddings.weight
727
+ module.decoder.layers.0.pre_mlp_layernorm.weight
728
+ module.decoder.layers.1.self_attention.linear_proj.weight
729
+ module.decoder.layers.0.self_attention.linear_qkv.layer_norm_weight
730
+ module.decoder.layers.0.self_attention.linear_proj.weight
731
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 20
732
+ Params for bucket 1 (41943040 elements, 41943040 padded size):
733
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight4
734
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight6
735
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight2
736
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight0
737
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight7
738
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight5
739
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight3
740
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight7
741
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight5
742
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight3
743
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight1
744
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight6
745
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight4
746
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight2
747
+ Params for bucket 2 (41943040 elements, 41943040 padded size):
748
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight1
749
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight5
750
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight7
751
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight3
752
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight1
753
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight6
754
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight4
755
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight6
756
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight0
757
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight4
758
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight2
759
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight0
760
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight7
761
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight5
762
+ Params for bucket 3 (41943040 elements, 41943040 padded size):
763
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight0
764
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight2
765
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight6
766
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight4
767
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight2
768
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight0
769
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight7
770
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight3
771
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight1
772
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight7
773
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight5
774
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight3
775
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight1
776
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight6
777
+ Params for bucket 4 (41943040 elements, 41943040 padded size):
778
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight5
779
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight3
780
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight1
781
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight7
782
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight5
783
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight3
784
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight1
785
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight2
786
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight4
787
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight0
788
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight6
789
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight4
790
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight2
791
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight0
792
+ Params for bucket 5 (41943040 elements, 41943040 padded size):
793
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight4
794
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight6
795
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight2
796
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight0
797
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight7
798
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight5
799
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight7
800
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight3
801
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight5
802
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight1
803
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight6
804
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight4
805
+ Params for bucket 6 (41943040 elements, 41943040 padded size):
806
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight3
807
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight1
808
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight6
809
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight4
810
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight2
811
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight0
812
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight0
813
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight7
814
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight2
815
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight5
816
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight3
817
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight1
818
+ Params for bucket 7 (41943040 elements, 41943040 padded size):
819
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight6
820
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight4
821
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight2
822
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight0
823
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight7
824
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight5
825
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight3
826
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight7
827
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight3
828
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight5
829
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight1
830
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight6
831
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight4
832
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight2
833
+ Params for bucket 8 (41943040 elements, 41943040 padded size):
834
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight1
835
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight5
836
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight7
837
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight3
838
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight1
839
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight6
840
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight4
841
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight6
842
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight0
843
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight4
844
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight2
845
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight0
846
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight7
847
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight5
848
+ Params for bucket 9 (41943040 elements, 41943040 padded size):
849
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight2
850
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight7
851
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight0
852
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight5
853
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight3
854
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight1
855
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight6
856
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight3
857
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight1
858
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight6
859
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight4
860
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight2
861
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight0
862
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight7
863
+ Params for bucket 10 (41943040 elements, 41943040 padded size):
864
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight4
865
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight2
866
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight0
867
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight6
868
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight4
869
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight2
870
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight0
871
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight5
872
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight1
873
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight3
874
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight7
875
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight5
876
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight3
877
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight1
878
+ Params for bucket 11 (41943040 elements, 41943040 padded size):
879
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight7
880
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight5
881
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight3
882
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight1
883
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight6
884
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight4
885
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight6
886
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight4
887
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight2
888
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight0
889
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight7
890
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight5
891
+ Params for bucket 12 (41943040 elements, 41943040 padded size):
892
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight0
893
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight2
894
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight6
895
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight4
896
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight2
897
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight0
898
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight3
899
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight7
900
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight1
901
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight5
902
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight3
903
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight1
904
+ Params for bucket 13 (41943040 elements, 41943040 padded size):
905
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight6
906
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight4
907
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight2
908
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight0
909
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight7
910
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight5
911
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight3
912
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight7
913
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight3
914
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight5
915
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight1
916
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight6
917
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight4
918
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight2
919
+ Params for bucket 14 (41943040 elements, 41943040 padded size):
920
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight1
921
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight5
922
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight7
923
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight3
924
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight1
925
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight6
926
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight4
927
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight0
928
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight6
929
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight4
930
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight2
931
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight0
932
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight7
933
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight5
934
+ Params for bucket 15 (41943040 elements, 41943040 padded size):
935
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight7
936
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight2
937
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight0
938
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight5
939
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight3
940
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight1
941
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight7
942
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight3
943
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight1
944
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight6
945
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight4
946
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight2
947
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight0
948
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight6
949
+ Params for bucket 16 (41943040 elements, 41943040 padded size):
950
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight5
951
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight1
952
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight3
953
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight6
954
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight4
955
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight2
956
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight0
957
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight2
958
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight4
959
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight0
960
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight7
961
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight5
962
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight3
963
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight1
964
+ Params for bucket 17 (41943040 elements, 41943040 padded size):
965
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight7
966
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight3
967
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight5
968
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight1
969
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight6
970
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight4
971
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight6
972
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight4
973
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight2
974
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight0
975
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight7
976
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight5
977
+ Params for bucket 18 (41943040 elements, 41943040 padded size):
978
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight0
979
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight2
980
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight6
981
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight4
982
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight2
983
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight0
984
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight3
985
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight1
986
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight7
987
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight5
988
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight3
989
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight1
990
+ Params for bucket 19 (41943040 elements, 41943040 padded size):
991
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight7
992
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight3
993
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight5
994
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight1
995
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight7
996
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight3
997
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight4
998
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight6
999
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight2
1000
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight0
1001
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight6
1002
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight5
1003
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight4
1004
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight2
1005
+ Params for bucket 20 (8388608 elements, 8388608 padded size):
1006
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight1
1007
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight0
1008
+ 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 0x7293c65c1d60>, config_logger_dir='')
1009
+ INFO:megatron.core.optimizer_param_scheduler:> learning rate decay style: cosine
1010
+ [after model, optimizer, and learning rate scheduler are built] datetime: 2026-09-21 08:33:39
1011
+ > building train, validation, and test datasets ...
1012
+ > datasets target sizes (minimum size):
1013
+ train: 96
1014
+ validation: 0
1015
+ test: 0
1016
+ INFO:megatron.core.datasets.blended_megatron_dataset_config:Let split_matrix = [(0, 1.0), None, None]
1017
+ > building train, validation, and test datasets for GPT ...
1018
+ INFO:megatron.core.datasets.blended_megatron_dataset_builder:Building GPTDataset splits with sizes=(96, 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 0x7293c7936840>, 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)
1019
+ INFO:megatron.core.datasets.indexed_dataset:Load the _IndexReader from /workspace/data/wikitext/wikitext103_gpt2_text_document.idx
1020
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence lengths
1021
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence pointers
1022
+ INFO:megatron.core.datasets.indexed_dataset: Extract the document indices
1023
+ INFO:megatron.core.datasets.indexed_dataset:> total number of sequences: 1165029
1024
+ INFO:megatron.core.datasets.indexed_dataset:> total number of documents: 1801350
1025
+ INFO:megatron.core.datasets.gpt_dataset:Load the GPTDataset train indices
1026
+ INFO:megatron.core.datasets.gpt_dataset: Load the document index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-document_index.npy
1027
+ INFO:megatron.core.datasets.gpt_dataset: Load the sample index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-sample_index.npy
1028
+ INFO:megatron.core.datasets.gpt_dataset: Load the shuffle index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-shuffle_index.npy
1029
+ INFO:megatron.core.datasets.gpt_dataset:> total number of samples: 29073
1030
+ > finished creating GPT datasets ...
1031
+ [after dataloaders are built] datetime: 2026-09-21 08:33:39
1032
+ done with setup ...
1033
+ (min, max) time across ranks (ms):
1034
+ model-and-optimizer-setup ......................: (293.06, 297.74)
1035
+ train/valid/test-data-iterators-setup ..........: (65.19, 70.97)
1036
+ training ...
1037
+ Overwriting rerun_state_machine.current_iteration from -1 to 0...
1038
+ [before the start of training step] datetime: 2026-09-21 08:33:39
1039
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1042
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1043
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1044
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1048
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1052
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1056
+ warnings.warn("The num_dispatched_tokens 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:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1062
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1063
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1066
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1067
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1070
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1071
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1072
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1073
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1074
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1075
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1076
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1077
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1078
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1079
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1080
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1081
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1082
+ warnings.warn("The num_dispatched_tokens 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:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1086
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1087
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1090
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1091
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1092
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1093
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1094
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1095
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1096
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1097
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1098
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1099
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1100
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1101
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1102
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1103
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1104
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1105
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1106
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1107
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1108
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1109
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1110
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1111
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1112
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1113
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1114
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1115
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1116
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1117
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1118
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1119
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1120
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1121
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1122
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1123
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1124
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1125
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1126
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1127
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1128
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1129
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1130
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1131
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1132
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1133
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1134
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1135
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1136
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1137
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1138
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1139
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1140
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1141
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1142
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1143
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1144
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1145
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1146
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1147
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1148
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1149
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1150
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1151
+ [2026-09-21 08:34:16] iteration 1/ 12 | consumed samples: 8 | elapsed time per iteration (ms): 36589.2 | throughput per GPU (TFLOP/s/GPU): 0.9 | learning rate: 2.000000E-05 | global batch size: 8 | lm loss: 1.118499E+01 | loss scale: 1.0 | grad norm: 35.477 | number of skipped iterations: 0 | number of nan iterations: 0 |Number of parameters in transformer block in billions: 6.71
1152
+
1153
+ Number of parameters in embedding layers in billions: 0.21
1154
+ Total number of parameters in billions: 6.92
1155
+ Number of parameters in most loaded shard in billions: 6.9171
1156
+ Theoretical memory footprints: weight and optimizer=49474.72 MB
1157
+ [Rank 0] (after 1 iterations) memory (MB) | allocated: 19780.87158203125 | max allocated: 23491.14599609375 | reserved: 24394.0 | max reserved: 24394.0
1158
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1159
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1160
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1161
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1162
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1163
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1164
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1165
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1166
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1167
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1168
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1169
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1170
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1171
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1172
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1173
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1174
+ [2026-09-21 08:34:16] iteration 2/ 12 | consumed samples: 16 | elapsed time per iteration (ms): 282.5 | throughput per GPU (TFLOP/s/GPU): 114.0 | learning rate: 4.000000E-05 | global batch size: 8 | lm loss: 1.120955E+01 | loss scale: 1.0 | grad norm: 36.285 | number of skipped iterations: 0 | number of nan iterations: 0 |
1175
+ [2026-09-21 08:34:17] iteration 3/ 12 | consumed samples: 24 | elapsed time per iteration (ms): 554.0 | throughput per GPU (TFLOP/s/GPU): 58.2 | learning rate: 6.000000E-05 | global batch size: 8 | lm loss: 1.006621E+01 | loss scale: 1.0 | grad norm: 101.491 | number of skipped iterations: 0 | number of nan iterations: 0 |
1176
+ Capture CUDA graph for training!!!
1177
+ CUDA graph capture done!!!
1178
+ [2026-09-21 08:34:18] iteration 4/ 12 | consumed samples: 32 | elapsed time per iteration (ms): 920.9 | throughput per GPU (TFLOP/s/GPU): 35.0 | learning rate: 8.000000E-05 | global batch size: 8 | lm loss: 1.023742E+01 | loss scale: 1.0 | grad norm: 29.934 | number of skipped iterations: 0 | number of nan iterations: 0 |
1179
+ [2026-09-21 08:34:18] iteration 5/ 12 | consumed samples: 40 | elapsed time per iteration (ms): 110.2 | throughput per GPU (TFLOP/s/GPU): 292.5 | learning rate: 1.000000E-04 | global batch size: 8 | lm loss: 1.077529E+01 | loss scale: 1.0 | grad norm: 21.350 | number of skipped iterations: 0 | number of nan iterations: 0 |
1180
+ [2026-09-21 08:34:18] iteration 6/ 12 | consumed samples: 48 | elapsed time per iteration (ms): 109.6 | throughput per GPU (TFLOP/s/GPU): 293.9 | learning rate: 7.750000E-05 | global batch size: 8 | lm loss: 9.671154E+00 | loss scale: 1.0 | grad norm: 8.514 | number of skipped iterations: 0 | number of nan iterations: 0 |
1181
+ [2026-09-21 08:34:18] iteration 7/ 12 | consumed samples: 56 | elapsed time per iteration (ms): 113.5 | throughput per GPU (TFLOP/s/GPU): 283.9 | learning rate: 3.250000E-05 | global batch size: 8 | lm loss: 9.130701E+00 | loss scale: 1.0 | grad norm: 15.214 | number of skipped iterations: 0 | number of nan iterations: 0 |
1182
+ [2026-09-21 08:34:18] iteration 8/ 12 | consumed samples: 64 | elapsed time per iteration (ms): 110.7 | throughput per GPU (TFLOP/s/GPU): 291.0 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.839638E+00 | loss scale: 1.0 | grad norm: 6.905 | number of skipped iterations: 0 | number of nan iterations: 0 |
1183
+ [2026-09-21 08:34:18] iteration 9/ 12 | consumed samples: 72 | elapsed time per iteration (ms): 111.3 | throughput per GPU (TFLOP/s/GPU): 289.3 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.723147E+00 | loss scale: 1.0 | grad norm: 4.123 | number of skipped iterations: 0 | number of nan iterations: 0 |
1184
+ [2026-09-21 08:34:18] iteration 10/ 12 | consumed samples: 80 | elapsed time per iteration (ms): 109.0 | throughput per GPU (TFLOP/s/GPU): 295.5 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.658899E+00 | loss scale: 1.0 | grad norm: 3.605 | number of skipped iterations: 0 | number of nan iterations: 0 |
1185
+ [2026-09-21 08:34:18] iteration 11/ 12 | consumed samples: 88 | elapsed time per iteration (ms): 108.4 | throughput per GPU (TFLOP/s/GPU): 297.1 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.643675E+00 | loss scale: 1.0 | grad norm: 3.522 | number of skipped iterations: 0 | number of nan iterations: 0 |
1186
+ [2026-09-21 08:34:19] iteration 12/ 12 | consumed samples: 96 | elapsed time per iteration (ms): 108.4 | throughput per GPU (TFLOP/s/GPU): 297.3 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.590739E+00 | loss scale: 1.0 | grad norm: 3.409 | number of skipped iterations: 0 | number of nan iterations: 0 |
1187
+ [after training is done] datetime: 2026-09-21 08:34:19
1188
+ [rank0]:[W921 08:34:20.219735171 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())
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2/mem_samples.csv ADDED
@@ -0,0 +1,224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 0, 0
2
+ 1, 0
3
+ 2, 0
4
+ 3, 0
5
+ 4, 0
6
+ 5, 0
7
+ 6, 0
8
+ 7, 0
9
+ 0, 0
10
+ 1, 0
11
+ 2, 0
12
+ 3, 0
13
+ 4, 0
14
+ 5, 0
15
+ 6, 0
16
+ 7, 0
17
+ 0, 4
18
+ 1, 1
19
+ 2, 1
20
+ 3, 1
21
+ 4, 1
22
+ 5, 1
23
+ 6, 1
24
+ 7, 1
25
+ 0, 4
26
+ 1, 4
27
+ 2, 4
28
+ 3, 4
29
+ 4, 1
30
+ 5, 1
31
+ 6, 1
32
+ 7, 1
33
+ 0, 4
34
+ 1, 4
35
+ 2, 4
36
+ 3, 4
37
+ 4, 4
38
+ 5, 4
39
+ 6, 4
40
+ 7, 4
41
+ 0, 821
42
+ 1, 821
43
+ 2, 821
44
+ 3, 821
45
+ 4, 821
46
+ 5, 821
47
+ 6, 821
48
+ 7, 821
49
+ 0, 1537
50
+ 1, 1537
51
+ 2, 1537
52
+ 3, 1537
53
+ 4, 1537
54
+ 5, 1537
55
+ 6, 1537
56
+ 7, 1537
57
+ 0, 1783
58
+ 1, 1875
59
+ 2, 1783
60
+ 3, 1783
61
+ 4, 1783
62
+ 5, 1783
63
+ 6, 1783
64
+ 7, 1783
65
+ 0, 13545
66
+ 1, 13549
67
+ 2, 13549
68
+ 3, 13549
69
+ 4, 13545
70
+ 5, 13549
71
+ 6, 13549
72
+ 7, 13545
73
+ 0, 14199
74
+ 1, 14201
75
+ 2, 14201
76
+ 3, 14203
77
+ 4, 14197
78
+ 5, 14201
79
+ 6, 14203
80
+ 7, 14197
81
+ 0, 14831
82
+ 1, 14833
83
+ 2, 14833
84
+ 3, 14835
85
+ 4, 14829
86
+ 5, 14833
87
+ 6, 14835
88
+ 7, 14829
89
+ 0, 14915
90
+ 1, 14917
91
+ 2, 14917
92
+ 3, 14919
93
+ 4, 14913
94
+ 5, 14917
95
+ 6, 14919
96
+ 7, 14913
97
+ 0, 15159
98
+ 1, 15173
99
+ 2, 15173
100
+ 3, 15175
101
+ 4, 15169
102
+ 5, 15157
103
+ 6, 15159
104
+ 7, 15151
105
+ 0, 15711
106
+ 1, 15713
107
+ 2, 15713
108
+ 3, 15715
109
+ 4, 15709
110
+ 5, 15713
111
+ 6, 15715
112
+ 7, 15709
113
+ 0, 15851
114
+ 1, 15853
115
+ 2, 15853
116
+ 3, 15855
117
+ 4, 15849
118
+ 5, 15853
119
+ 6, 15855
120
+ 7, 15849
121
+ 0, 16349
122
+ 1, 16351
123
+ 2, 16351
124
+ 3, 16353
125
+ 4, 16347
126
+ 5, 16351
127
+ 6, 16351
128
+ 7, 16347
129
+ 0, 16349
130
+ 1, 16351
131
+ 2, 16351
132
+ 3, 16353
133
+ 4, 16347
134
+ 5, 16351
135
+ 6, 16351
136
+ 7, 16347
137
+ 0, 28963
138
+ 1, 28965
139
+ 2, 28975
140
+ 3, 28967
141
+ 4, 28961
142
+ 5, 28965
143
+ 6, 28967
144
+ 7, 28961
145
+ 0, 28963
146
+ 1, 28965
147
+ 2, 28975
148
+ 3, 28967
149
+ 4, 28961
150
+ 5, 28965
151
+ 6, 28967
152
+ 7, 28961
153
+ 0, 28963
154
+ 1, 28965
155
+ 2, 28975
156
+ 3, 28967
157
+ 4, 28961
158
+ 5, 28965
159
+ 6, 28967
160
+ 7, 28961
161
+ 0, 28963
162
+ 1, 28965
163
+ 2, 28975
164
+ 3, 28967
165
+ 4, 28961
166
+ 5, 28965
167
+ 6, 28967
168
+ 7, 28961
169
+ 0, 28963
170
+ 1, 28965
171
+ 2, 28975
172
+ 3, 28967
173
+ 4, 28961
174
+ 5, 28965
175
+ 6, 28967
176
+ 7, 28961
177
+ 0, 30281
178
+ 1, 29631
179
+ 2, 30285
180
+ 3, 29641
181
+ 4, 29645
182
+ 5, 30299
183
+ 6, 30315
184
+ 7, 30283
185
+ 0, 31129
186
+ 1, 30453
187
+ 2, 31101
188
+ 3, 30461
189
+ 4, 30473
190
+ 5, 31163
191
+ 6, 31147
192
+ 7, 31117
193
+ 0, 38499
194
+ 1, 38003
195
+ 2, 38527
196
+ 3, 37989
197
+ 4, 38189
198
+ 5, 38517
199
+ 6, 38499
200
+ 7, 38509
201
+ 0, 42895
202
+ 1, 43043
203
+ 2, 42911
204
+ 3, 42905
205
+ 4, 43035
206
+ 5, 42903
207
+ 6, 42905
208
+ 7, 43035
209
+ 0, 42099
210
+ 1, 42247
211
+ 2, 42115
212
+ 3, 42109
213
+ 4, 2102
214
+ 5, 42107
215
+ 6, 2102
216
+ 7, 42239
217
+ 0, 0
218
+ 1, 0
219
+ 2, 0
220
+ 3, 0
221
+ 4, 0
222
+ 5, 0
223
+ 6, 0
224
+ 7, 0
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2/run.env ADDED
@@ -0,0 +1,102 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ backend=echo
2
+ config=OLMOE_1B_7B_L16
3
+ family=tierB
4
+ source_model=OLMoE-1B-7B-0924
5
+ standard_num_layers=16
6
+ base_rung=
7
+ repo=/workspace/Megatron-MoE-Echo
8
+ branch=bracing-regions-through-kernels
9
+ commit=9a236793
10
+ num_layers=16
11
+ hidden=2048
12
+ ffn_hidden=1024
13
+ moe_ffn_hidden=1024
14
+ num_experts=64
15
+ topk=8
16
+ experts_per_rank=8
17
+ expert_params=6291456
18
+ heads=16
19
+ kv_groups=16
20
+ kv_channels=128
21
+ qk_layernorm=0
22
+ shared_ffn=0
23
+ shared_expert_overlap=0
24
+ moe_token_dispatcher=flex
25
+ mla=0
26
+ q_lora_rank=
27
+ kv_lora_rank=512
28
+ qk_head_dim=128
29
+ qk_pos_emb_head_dim=64
30
+ v_head_dim=128
31
+ dense_first_layers=0
32
+ moe_layers=16
33
+ rotary_base=10000
34
+ rotary_percent=1.0
35
+ rotary_scaling=
36
+ mscale=
37
+ mscale_all_dim=
38
+ qkv_bias=0
39
+ router_balancing=aux_loss
40
+ aux_loss_coeff=1e-2
41
+ router_pre_softmax=1
42
+ router_fusion=0
43
+ router_dtype=fp32
44
+ router_groups=
45
+ router_group_topk=
46
+ router_score_function=softmax
47
+ router_expert_bias=0
48
+ router_bias_update_rate=
49
+ router_topk_scaling_factor=
50
+ shared_expert_gate=0
51
+ apply_layernorm_1p=0
52
+ attention_output_gate=0
53
+ linear_attention_type=
54
+ linear_attention_freq=
55
+ linear_conv_kernel_dim=
56
+ linear_key_head_dim=
57
+ linear_value_head_dim=
58
+ linear_num_key_heads=
59
+ linear_num_value_heads=
60
+ mtp_num_layers=0
61
+ mtp_loss_scaling_factor=0.1
62
+ no_weight_decay_cond_type=
63
+ max_position_embeddings=4096
64
+ init_method_std=0.02
65
+ device_initiated_grouped_gemm=1
66
+ norm_epsilon=1e-5
67
+ seq_length=4096
68
+ micro_batch_size=1
69
+ global_batch_size=8
70
+ num_microbatches=1
71
+ assignments_per_rank=32768
72
+ mean_tokens_per_expert=4096
73
+ slot_policy=matched
74
+ slots_per_rank=8
75
+ capacity_mode=static
76
+ capacity=1.08984375
77
+ cuda_graph=1
78
+ probe_mode=0
79
+ async_grad_reduce=0
80
+ branch_role=timing
81
+ memory_dump_mode=none
82
+ memory_snapshot_path=
83
+ cuda_graph_warmup_steps=3
84
+ recompute=none
85
+ force_balance=0
86
+ train_iters=12
87
+ ep_size=8
88
+ echo_dispatch_overlap=1
89
+ echo_dispatch_recompute=0
90
+ ultraep_quota_min_tokens=
91
+ ultraep_weight_sync_join=
92
+ profile_enable=0
93
+ kernel_trace=0
94
+ fp8_mode=mxfp8
95
+ fp8_recipe=mxfp8
96
+ training_seed=1234
97
+ learning_rate=1e-4
98
+ min_learning_rate=1.0e-5
99
+ lr_warmup_iters=5
100
+ state_dump_enable=0
101
+ state_dump_dir=
102
+ moonep_assert_no_pending_reduce=0
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2/train.log ADDED
@@ -0,0 +1,1188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==============================================================
2
+ OLMOE_1B_7B_L16 / echo (slots/rank=8 policy=matched)
3
+ E=64 K=8 H'=1024 m=8
4
+ L=16 seq=4096 gbs=8 (1 microbatches)
5
+ cap_mode=static cap=1.08984375 cg=1 recompute=none
6
+ memory_dump=none snapshots=/workspace/granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2/memory_snapshots/snapshot.pickle
7
+ repo=/workspace/Megatron-MoE-Echo @ 9a236793
8
+ ==============================================================
9
+ W0921 08:34:27.062000 683457 torch/distributed/run.py:766]
10
+ W0921 08:34:27.062000 683457 torch/distributed/run.py:766] *****************************************
11
+ W0921 08:34:27.062000 683457 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
+ W0921 08:34:27.062000 683457 torch/distributed/run.py:766] *****************************************
13
+ /workspace/Megatron-MoE-Echo/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-MoE-Echo/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-MoE-Echo/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
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
20
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
21
+ 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
22
+ WARNING: overriding default arguments for tokenizer_type:GPT2BPETokenizer with tokenizer_type:GPT2BPETokenizer
23
+ Number of virtual stages per pipeline stage: None
24
+ accumulate and all-reduce gradients in fp32 for bfloat16 data type.
25
+ using torch.bfloat16 for parameters ...
26
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
27
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
28
+ ------------------------ arguments ------------------------
29
+ account_for_embedding_in_pipeline_split ......... False
30
+ account_for_loss_in_pipeline_split .............. False
31
+ accumulate_allreduce_grads_in_fp32 .............. True
32
+ adam_beta1 ...................................... 0.9
33
+ adam_beta2 ...................................... 0.999
34
+ adam_eps ........................................ 1e-08
35
+ add_bias_linear ................................. False
36
+ add_position_embedding .......................... False
37
+ add_qkv_bias .................................... False
38
+ adlr_autoresume ................................. False
39
+ adlr_autoresume_interval ........................ 1000
40
+ align_grad_reduce ............................... True
41
+ align_param_gather .............................. False
42
+ app_tag_run_name ................................ None
43
+ app_tag_run_version ............................. 0.0.0
44
+ apply_layernorm_1p .............................. False
45
+ apply_query_key_layer_scaling ................... False
46
+ apply_residual_connection_post_layernorm ........ False
47
+ apply_rope_fusion ............................... True
48
+ async_save ...................................... None
49
+ async_tensor_model_parallel_allreduce ........... True
50
+ attention_backend ............................... AttnBackend.auto
51
+ attention_dropout ............................... 0.0
52
+ attention_softmax_in_fp32 ....................... False
53
+ auto_detect_ckpt_format ......................... False
54
+ barrier_with_L1_time ............................ True
55
+ bert_binary_head ................................ True
56
+ bert_embedder_type .............................. megatron
57
+ bert_load ....................................... None
58
+ bf16 ............................................ True
59
+ bias_dropout_fusion ............................. True
60
+ bias_gelu_fusion ................................ False
61
+ bias_swiglu_fusion .............................. True
62
+ biencoder_projection_dim ........................ 0
63
+ biencoder_shared_query_context_model ............ False
64
+ block_data_path ................................. None
65
+ calc_ft_timeouts ................................ False
66
+ calculate_per_token_loss ........................ False
67
+ check_for_large_grads ........................... False
68
+ check_for_nan_in_loss_and_grad .................. False
69
+ check_for_spiky_loss ............................ False
70
+ check_weight_hash_across_dp_replicas_interval ... None
71
+ ckpt_assume_constant_structure .................. False
72
+ ckpt_convert_format ............................. None
73
+ ckpt_convert_save ............................... None
74
+ ckpt_convert_update_legacy_dist_opt_format ...... False
75
+ ckpt_format ..................................... torch_dist
76
+ ckpt_fully_parallel_load ........................ False
77
+ ckpt_fully_parallel_save ........................ True
78
+ ckpt_fully_parallel_save_deprecated ............. False
79
+ ckpt_step ....................................... None
80
+ classes_fraction ................................ 1.0
81
+ clip_grad ....................................... 1.0
82
+ clone_scatter_output_in_embedding ............... True
83
+ config_logger_dir ...............................
84
+ consumed_train_samples .......................... 0
85
+ consumed_valid_samples .......................... 0
86
+ context_parallel_size ........................... 1
87
+ cp_comm_type .................................... ['p2p']
88
+ create_attention_mask_in_dataloader ............. True
89
+ cross_entropy_fusion_impl ....................... native
90
+ cross_entropy_loss_fusion ....................... False
91
+ cuda_graph_scope ................................ full_iteration
92
+ cuda_graph_warmup_steps ......................... 3
93
+ data_args_path .................................. None
94
+ data_cache_path ................................. None
95
+ data_parallel_random_init ....................... False
96
+ data_parallel_sharding_strategy ................. no_shard
97
+ data_parallel_size .............................. 8
98
+ data_path ....................................... ['/workspace/data/wikitext/wikitext103_gpt2_text_document']
99
+ data_per_class_fraction ......................... 1.0
100
+ data_sharding ................................... True
101
+ dataloader_type ................................. single
102
+ ddp_average_in_collective ....................... False
103
+ ddp_bucket_size ................................. None
104
+ ddp_num_buckets ................................. None
105
+ ddp_pad_buckets_for_high_nccl_busbw ............. False
106
+ decoder_first_pipeline_num_layers ............... None
107
+ decoder_last_pipeline_num_layers ................ None
108
+ decoder_num_layers .............................. None
109
+ decoder_seq_length .............................. None
110
+ decoupled_lr .................................... None
111
+ decoupled_min_lr ................................ None
112
+ decrease_batch_size_if_needed ................... False
113
+ defer_embedding_wgrad_compute ................... False
114
+ delay_wgrad_compute ............................. False
115
+ deprecated_use_mcore_models ..................... True
116
+ deterministic_mode .............................. False
117
+ dino_bottleneck_size ............................ 256
118
+ dino_freeze_last_layer .......................... 1
119
+ dino_head_hidden_size ........................... 2048
120
+ dino_local_crops_number ......................... 10
121
+ dino_local_img_size ............................. 96
122
+ dino_norm_last_layer ............................ False
123
+ dino_teacher_temp ............................... 0.07
124
+ dino_warmup_teacher_temp ........................ 0.04
125
+ dino_warmup_teacher_temp_epochs ................. 30
126
+ disable_bf16_reduced_precision_matmul ........... False
127
+ disable_mamba_mem_eff_path ...................... False
128
+ disable_straggler_on_startup .................... False
129
+ dist_ckpt_format_deprecated ..................... None
130
+ dist_ckpt_strictness ............................ assume_ok_unexpected
131
+ distribute_saved_activations .................... False
132
+ distributed_backend ............................. nccl
133
+ distributed_timeout_minutes ..................... 10
134
+ embedding_init_method_std ....................... None
135
+ embedding_path .................................. None
136
+ empty_unused_memory_level ....................... 0
137
+ enable_cuda_graph ............................... True
138
+ enable_experimental ............................. False
139
+ enable_ft_package ............................... False
140
+ enable_gloo_process_groups ...................... True
141
+ enable_msc ...................................... True
142
+ enable_one_logger ............................... True
143
+ encoder_num_layers .............................. 16
144
+ encoder_seq_length .............................. 4096
145
+ end_weight_decay ................................ 0.1
146
+ eod_mask_loss ................................... False
147
+ error_injection_rate ............................ 0
148
+ error_injection_type ............................ transient_error
149
+ eval_interval ................................... 1000000
150
+ eval_iters ...................................... 0
151
+ evidence_data_path .............................. None
152
+ exit_duration_in_mins ........................... None
153
+ exit_interval ................................... None
154
+ exit_on_missing_checkpoint ...................... False
155
+ exit_signal_handler ............................. False
156
+ exp_avg_dtype ................................... torch.float32
157
+ exp_avg_sq_dtype ................................ torch.float32
158
+ expert_model_parallel_size ...................... 8
159
+ expert_tensor_parallel_size ..................... 1
160
+ export_force_local_attention .................... False
161
+ export_kd_cfg ................................... None
162
+ export_kd_teacher_ckpt_format ................... None
163
+ export_kd_teacher_load .......................... None
164
+ export_kv_cache_quant ........................... False
165
+ export_legacy_megatron .......................... False
166
+ export_model_type ............................... GPTModel
167
+ export_moe_apply_probs_on_input ................. False
168
+ export_qk_l2_norm ............................... False
169
+ export_quant_cfg ................................ None
170
+ export_real_quant_cfg ........................... None
171
+ export_te_mcore_model ........................... False
172
+ external_cuda_graph ............................. False
173
+ ffn_hidden_size ................................. 1024
174
+ finetune ........................................ False
175
+ finetune_data_split ............................. train
176
+ finetune_hf_dataset ............................. None
177
+ first_last_layers_bf16 .......................... False
178
+ flash_decode .................................... False
179
+ fp16 ............................................ False
180
+ fp16_lm_cross_entropy ........................... False
181
+ fp32_residual_connection ........................ False
182
+ fp8 ............................................. e4m3
183
+ fp8_amax_compute_algo ........................... most_recent
184
+ fp8_amax_history_len ............................ 1
185
+ fp8_interval .................................... 1
186
+ fp8_margin ...................................... 0
187
+ fp8_param_gather ................................ False
188
+ fp8_recipe ...................................... mxfp8
189
+ fp8_wgrad ....................................... True
190
+ fsdp_double_buffer .............................. False
191
+ global_batch_size ............................... 8
192
+ grad_reduce_in_bf16 ............................. False
193
+ gradient_accumulation_fusion .................... True
194
+ gradient_reduce_div_fusion ...................... True
195
+ group_query_attention ........................... True
196
+ head_lr_mult .................................... 1.0
197
+ heterogeneous_layers_config_encoded_json ........ None
198
+ heterogeneous_layers_config_path ................ None
199
+ hidden_dropout .................................. 0.0
200
+ hidden_size ..................................... 2048
201
+ hierarchical_context_parallel_sizes ............. None
202
+ high_priority_stream_groups ..................... []
203
+ hybrid_attention_ratio .......................... 0.0
204
+ hybrid_mlp_ratio ................................ 0.0
205
+ hybrid_override_pattern ......................... None
206
+ hysteresis ...................................... 2
207
+ ict_head_size ................................... None
208
+ ict_load ........................................ None
209
+ img_h ........................................... 224
210
+ img_w ........................................... 224
211
+ indexer_batch_size .............................. 128
212
+ indexer_log_interval ............................ 1000
213
+ inference_batch_times_seqlen_threshold .......... -1
214
+ inference_dynamic_batching ...................... False
215
+ inference_dynamic_batching_buffer_guaranteed_fraction 0.2
216
+ inference_dynamic_batching_buffer_overflow_factor None
217
+ inference_dynamic_batching_buffer_size_gb ....... 40.0
218
+ inference_dynamic_batching_chunk_size ........... 256
219
+ inference_dynamic_batching_max_requests_override None
220
+ inference_dynamic_batching_max_tokens_override .. None
221
+ inference_dynamic_batching_num_cuda_graphs ...... 16
222
+ inference_max_batch_size ........................ 8
223
+ inference_max_seq_length ........................ 2560
224
+ inference_rng_tracker ........................... False
225
+ init_method_std ................................. 0.02
226
+ init_method_xavier_uniform ...................... False
227
+ init_model_with_meta_device ..................... False
228
+ initial_loss_scale .............................. 4294967296
229
+ inprocess_active_world_size ..................... 8
230
+ inprocess_barrier_timeout ....................... 120
231
+ inprocess_completion_timeout .................... 120
232
+ inprocess_empty_cuda_cache ...................... False
233
+ inprocess_granularity ........................... node
234
+ inprocess_hard_timeout .......................... 90
235
+ inprocess_heartbeat_interval .................... 30
236
+ inprocess_heartbeat_timeout ..................... 60
237
+ inprocess_last_call_wait ........................ 1
238
+ inprocess_max_iterations ........................ None
239
+ inprocess_monitor_process_interval .............. 1.0
240
+ inprocess_monitor_thread_interval ............... 1.0
241
+ inprocess_progress_watchdog_interval ............ 1.0
242
+ inprocess_restart ............................... False
243
+ inprocess_soft_timeout .......................... 60
244
+ inprocess_termination_grace_time ................ 1
245
+ is_hybrid_model ................................. False
246
+ iter_per_epoch .................................. 1250
247
+ iterations_to_skip .............................. []
248
+ keep_fp8_transpose_cache_when_using_custom_fsdp . False
249
+ kitchen_config_file ............................. None
250
+ kitchen_recipe_number ........................... None
251
+ kv_channels ..................................... 128
252
+ kv_lora_rank .................................... 32
253
+ lazy_mpu_init ................................... None
254
+ load ............................................ None
255
+ load_main_params_from_ckpt ...................... None
256
+ load_model_opt_format ........................... False
257
+ local_rank ...................................... 0
258
+ log_energy ...................................... False
259
+ log_interval .................................... 1
260
+ log_loss_scale_to_tensorboard ................... True
261
+ log_memory_to_tensorboard ....................... False
262
+ log_num_zeros_in_grad ........................... False
263
+ log_params_norm ................................. False
264
+ log_progress .................................... False
265
+ log_straggler ................................... False
266
+ log_throughput .................................. True
267
+ log_timers_to_tensorboard ....................... False
268
+ log_validation_ppl_to_tensorboard ............... False
269
+ log_world_size_to_tensorboard ................... False
270
+ logging_level ................................... None
271
+ loss_scale ...................................... None
272
+ loss_scale_window ............................... 1000
273
+ lr .............................................. 0.0001
274
+ lr_decay_iters .................................. 8
275
+ lr_decay_samples ................................ None
276
+ lr_decay_style .................................. cosine
277
+ lr_warmup_fraction .............................. None
278
+ lr_warmup_init .................................. 0.0
279
+ lr_warmup_iters ................................. 5
280
+ lr_warmup_samples ............................... 0
281
+ lr_wsd_decay_iters .............................. None
282
+ lr_wsd_decay_samples ............................ None
283
+ lr_wsd_decay_style .............................. exponential
284
+ main_grads_dtype ................................ torch.float32
285
+ main_params_dtype ............................... torch.float32
286
+ make_vocab_size_divisible_by .................... 128
287
+ mamba_head_dim .................................. 64
288
+ mamba_num_groups ................................ 8
289
+ mamba_num_heads ................................. None
290
+ mamba_state_dim ................................. 128
291
+ manual_gc ....................................... False
292
+ manual_gc_eval .................................. True
293
+ manual_gc_interval .............................. 0
294
+ mask_factor ..................................... 1.0
295
+ mask_prob ....................................... 0.15
296
+ mask_type ....................................... random
297
+ masked_softmax_fusion ........................... False
298
+ max_position_embeddings ......................... 4096
299
+ max_tokens_to_oom ............................... 12000
300
+ memory_snapshot_path ............................ snapshot.pickle
301
+ merge_file ...................................... /workspace/data/wikitext/gpt2-merges.txt
302
+ micro_batch_size ................................ 1
303
+ microbatch_group_size_per_vp_stage .............. None
304
+ mid_level_dataset_surplus ....................... 0.005
305
+ min_loss_scale .................................. 1.0
306
+ min_lr .......................................... 1e-05
307
+ mlp_chunks_for_prefill .......................... 1
308
+ mmap_bin_files .................................. True
309
+ mock_data ....................................... False
310
+ moe_apply_probs_on_input ........................ False
311
+ moe_aux_loss_coeff .............................. 0.01
312
+ moe_deepep_num_sms .............................. 20
313
+ moe_echo_enable_random_offloading ............... False
314
+ moe_echo_expert_dispatch_overlap ................ True
315
+ moe_echo_expert_dispatcher_type ................. hybridep
316
+ moe_echo_recompute_expert_dispatch .............. False
317
+ moe_enable_deepep ............................... False
318
+ moe_enable_echo ................................. True
319
+ moe_enable_hybridep ............................. True
320
+ moe_expert_capacity_factor ...................... None
321
+ moe_extended_tp ................................. False
322
+ moe_ffn_hidden_size ............................. 1024
323
+ moe_grouped_gemm ................................ True
324
+ moe_hybridep_num_sms ............................ 16
325
+ moe_input_jitter_eps ............................ None
326
+ moe_layer_freq .................................. 1
327
+ moe_layer_recompute ............................. False
328
+ moe_num_echo_experts ............................ 64
329
+ moe_pad_expert_input_to_capacity ................ False
330
+ moe_per_layer_logging ........................... False
331
+ moe_permute_fusion .............................. False
332
+ moe_received_token_capacity ..................... 1.08984375
333
+ moe_router_bias_update_rate ..................... 0.001
334
+ moe_router_dtype ................................ fp32
335
+ moe_router_enable_expert_bias ................... False
336
+ moe_router_force_load_balancing ................. False
337
+ moe_router_group_topk ........................... None
338
+ moe_router_load_balancing_type .................. aux_loss
339
+ moe_router_num_groups ........................... None
340
+ moe_router_padding_for_fp8 ...................... False
341
+ moe_router_pre_softmax .......................... True
342
+ moe_router_score_function ....................... softmax
343
+ moe_router_topk ................................. 8
344
+ moe_router_topk_scaling_factor .................. None
345
+ moe_shared_expert_intermediate_size ............. None
346
+ moe_shared_expert_overlap ....................... False
347
+ moe_token_dispatcher_type ....................... flex
348
+ moe_token_drop_policy ........................... probs
349
+ moe_upcycling_granularity ....................... 1
350
+ moe_use_device_initiated_grouped_gemm ........... True
351
+ moe_use_legacy_grouped_gemm ..................... False
352
+ moe_use_upcycling ............................... False
353
+ moe_z_loss_coeff ................................ None
354
+ mrope_section ................................... None
355
+ mscale .......................................... 1.0
356
+ mscale_all_dim .................................. 0.0
357
+ mtp_loss_scaling_factor ......................... 0.1
358
+ mtp_num_layers .................................. None
359
+ multi_latent_attention .......................... False
360
+ nccl_all_reduce_for_prefill ..................... False
361
+ nccl_communicator_config_path ................... None
362
+ nccl_ub ......................................... False
363
+ no_load_optim ................................... None
364
+ no_load_rng ..................................... None
365
+ no_persist_layer_norm ........................... False
366
+ no_rope_freq .................................... None
367
+ no_save_optim ................................... None
368
+ no_save_rng ..................................... None
369
+ non_persistent_ckpt_type ........................ None
370
+ non_persistent_global_ckpt_dir .................. None
371
+ non_persistent_local_ckpt_algo .................. fully_parallel
372
+ non_persistent_local_ckpt_dir ................... None
373
+ non_persistent_save_interval .................... None
374
+ norm_epsilon .................................... 1e-05
375
+ normalization ................................... RMSNorm
376
+ num_attention_heads ............................. 16
377
+ num_channels .................................... 3
378
+ num_classes ..................................... 1000
379
+ num_dataset_builder_threads ..................... 1
380
+ num_distributed_optimizer_instances ............. 1
381
+ num_experts ..................................... 64
382
+ num_layers ...................................... 16
383
+ num_layers_at_end_in_bf16 ....................... 1
384
+ num_layers_at_start_in_bf16 ..................... 1
385
+ num_layers_per_virtual_pipeline_stage ........... None
386
+ num_query_groups ................................ 16
387
+ num_virtual_stages_per_pipeline_rank ............ None
388
+ num_workers ..................................... 2
389
+ object_storage_cache_path ....................... None
390
+ one_logger_async ................................ False
391
+ one_logger_project .............................. megatron-lm
392
+ one_logger_run_name ............................. None
393
+ onnx_safe ....................................... None
394
+ openai_gelu ..................................... False
395
+ optimizer ....................................... adam
396
+ optimizer_cpu_offload ........................... False
397
+ optimizer_offload_fraction ...................... 1.0
398
+ output_bert_embeddings .......................... False
399
+ overlap_cpu_optimizer_d2h_h2d ................... False
400
+ overlap_grad_reduce ............................. True
401
+ overlap_moe_expert_parallel_comm ................ False
402
+ overlap_p2p_comm ................................ False
403
+ overlap_p2p_comm_warmup_flush ................... False
404
+ overlap_param_gather ............................ True
405
+ overlap_param_gather_with_optimizer_step ........ False
406
+ override_opt_param_scheduler .................... False
407
+ params_dtype .................................... torch.bfloat16
408
+ patch_dim ....................................... 16
409
+ per_split_data_args_path ........................ None
410
+ perform_initialization .......................... True
411
+ pin_cpu_grads ................................... True
412
+ pin_cpu_params .................................. True
413
+ pipeline_model_parallel_comm_backend ............ None
414
+ pipeline_model_parallel_layout .................. None
415
+ pipeline_model_parallel_size .................... 1
416
+ position_embedding_type ......................... rope
417
+ pretrained_checkpoint ........................... None
418
+ profile ......................................... False
419
+ profile_ranks ................................... [0]
420
+ profile_step_end ................................ 12
421
+ profile_step_start .............................. 10
422
+ q_lora_rank ..................................... None
423
+ qk_head_dim ..................................... 128
424
+ qk_l2_norm ...................................... False
425
+ qk_layernorm .................................... False
426
+ qk_pos_emb_head_dim ............................. 64
427
+ query_in_block_prob ............................. 0.1
428
+ rampup_batch_size ............................... None
429
+ rank ............................................ 0
430
+ recompute_granularity ........................... None
431
+ recompute_method ................................ None
432
+ recompute_modules ............................... None
433
+ recompute_num_layers ............................ None
434
+ record_memory_history ........................... False
435
+ relative_attention_max_distance ................. 128
436
+ relative_attention_num_buckets .................. 32
437
+ replication ..................................... False
438
+ replication_factor .............................. 2
439
+ replication_jump ................................ None
440
+ rerun_mode ...................................... validate_results
441
+ reset_attention_mask ............................ False
442
+ reset_position_ids .............................. False
443
+ result_rejected_tracker_filename ................ None
444
+ retriever_report_topk_accuracies ................ []
445
+ retriever_score_scaling ......................... False
446
+ retriever_seq_length ............................ 256
447
+ retro_add_retriever ............................. False
448
+ retro_attention_gate ............................ 1
449
+ retro_cyclic_train_iters ........................ None
450
+ retro_encoder_attention_dropout ................. 0.1
451
+ retro_encoder_hidden_dropout .................... 0.1
452
+ retro_encoder_layers ............................ 2
453
+ retro_num_neighbors ............................. 2
454
+ retro_num_retrieved_chunks ...................... 2
455
+ retro_project_dir ............................... None
456
+ retro_verify_neighbor_count ..................... True
457
+ reuse_grad_buf_for_mxfp8_param_ag ............... False
458
+ rope_scaling_factor ............................. 8.0
459
+ rope_type ....................................... None
460
+ rotary_base ..................................... 10000
461
+ rotary_interleaved .............................. False
462
+ rotary_percent .................................. 1.0
463
+ rotary_scaling_factor ........................... 1.0
464
+ rotary_seq_len_interpolation_factor ............. None
465
+ run_workload_inspector_server ................... False
466
+ sample_rate ..................................... 1.0
467
+ save ............................................ None
468
+ save_interval ................................... None
469
+ save_retain_interval ............................ None
470
+ scatter_gather_tensors_in_pipeline .............. True
471
+ seed ............................................ 1234
472
+ seq_length ...................................... 4096
473
+ sequence_parallel ............................... False
474
+ sft ............................................. False
475
+ sft_tokenizer_prompt_format ..................... nemotron-h-aligned
476
+ sgd_momentum .................................... 0.9
477
+ short_seq_prob .................................. 0.1
478
+ skip_train ...................................... False
479
+ skipped_train_samples ........................... 0
480
+ spec ............................................ None
481
+ split ........................................... 100,0,0
482
+ squared_relu .................................... False
483
+ start_weight_decay .............................. 0.1
484
+ straggler_ctrlr_port ............................ 65535
485
+ straggler_minmax_count .......................... 1
486
+ suggested_communication_unit_size ............... None
487
+ swiglu .......................................... True
488
+ swin_backbone_type .............................. tiny
489
+ symmetric_ar_type ............................... None
490
+ te_rng_tracker .................................. True
491
+ tensor_model_parallel_size ...................... 1
492
+ tensorboard_dir ................................. None
493
+ tensorboard_log_interval ........................ 1
494
+ tensorboard_queue_size .......................... 1000
495
+ test_data_path .................................. None
496
+ test_mode ....................................... False
497
+ tiktoken_num_special_tokens ..................... 1000
498
+ tiktoken_pattern ................................ None
499
+ tiktoken_special_tokens ......................... None
500
+ timing_log_level ................................ 0
501
+ timing_log_option ............................... minmax
502
+ titles_data_path ................................ None
503
+ tokenizer_model ................................. None
504
+ tokenizer_type .................................. GPT2BPETokenizer
505
+ torch_fsdp2_reshard_after_forward ............... True
506
+ tp_comm_bootstrap_backend ....................... nccl
507
+ tp_comm_bulk_dgrad .............................. True
508
+ tp_comm_bulk_wgrad .............................. True
509
+ tp_comm_overlap ................................. False
510
+ tp_comm_overlap_ag .............................. True
511
+ tp_comm_overlap_cfg ............................. None
512
+ tp_comm_overlap_rs .............................. True
513
+ tp_comm_overlap_rs_dgrad ........................ False
514
+ tp_comm_split_ag ................................ True
515
+ tp_comm_split_rs ................................ True
516
+ train_data_path ................................. None
517
+ train_iters ..................................... 12
518
+ train_samples ................................... None
519
+ train_sync_interval ............................. None
520
+ transformer_impl ................................ transformer_engine
521
+ transformer_pipeline_model_parallel_size ........ 1
522
+ untie_embeddings_and_output_weights ............. True
523
+ use_checkpoint_args ............................. False
524
+ use_checkpoint_opt_param_scheduler .............. False
525
+ use_cpu_initialization .......................... None
526
+ use_custom_fsdp ................................. False
527
+ use_dist_ckpt ................................... True
528
+ use_dist_ckpt_deprecated ........................ False
529
+ use_distributed_optimizer ....................... True
530
+ use_flash_attn .................................. False
531
+ use_legacy_models ............................... False
532
+ use_mp_args_from_checkpoint_args ................ False
533
+ use_one_sent_docs ............................... False
534
+ use_persistent_ckpt_worker ...................... False
535
+ use_precision_aware_optimizer ................... False
536
+ use_pytorch_profiler ............................ False
537
+ use_ring_exchange_p2p ........................... False
538
+ use_rope_scaling ................................ False
539
+ use_rotary_position_embeddings .................. False
540
+ use_sharp ....................................... False
541
+ use_tokenizer_model_from_checkpoint_args ........ True
542
+ use_torch_fsdp2 ................................. False
543
+ use_torch_optimizer_for_cpu_offload ............. False
544
+ use_tp_pp_dp_mapping ............................ False
545
+ v_head_dim ...................................... 128
546
+ valid_data_path ................................. None
547
+ variable_seq_lengths ............................ False
548
+ virtual_pipeline_model_parallel_size ............ None
549
+ vision_backbone_type ............................ vit
550
+ vision_pretraining .............................. False
551
+ vision_pretraining_type ......................... classify
552
+ vocab_extra_ids ................................. 0
553
+ vocab_file ...................................... /workspace/data/wikitext/gpt2-vocab.json
554
+ vocab_size ...................................... None
555
+ wandb_exp_name ..................................
556
+ wandb_project ...................................
557
+ wandb_save_dir ..................................
558
+ weight_decay .................................... 0.1
559
+ weight_decay_incr_style ......................... constant
560
+ wgrad_deferral_limit ............................ 0
561
+ world_size ...................................... 8
562
+ yaml_cfg ........................................ None
563
+ -------------------- end of arguments ---------------------
564
+ INFO:megatron.core.num_microbatches_calculator:setting number of microbatches to constant 1
565
+ > building GPT2BPETokenizer tokenizer ...
566
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
567
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
568
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
569
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
570
+ > padded vocab (size: 50257) with 47 dummy tokens (new size: 50304)
571
+ WARNING:megatron.core.rerun_state_machine:RerunStateMachine initialized in mode RerunMode.VALIDATE_RESULTS
572
+ > initializing torch distributed ...
573
+ 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
574
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
575
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
576
+ [Gloo] Rank [Gloo] Rank 0[Gloo] Rank [Gloo] Rank [Gloo] Rank [Gloo] Rank 1 is connected to 2743 is connected to 7 is connected to is connected to is connected to is connected to 7 peer ranks. 7777 peer ranks. Expected number of connected peer ranks is : peer ranks. peer ranks. peer ranks. peer ranks. Expected number of connected peer ranks is : 7Expected number of connected peer ranks is : Expected number of connected peer ranks is : Expected number of connected peer ranks is : Expected number of connected peer ranks is : 7
577
+ 777
578
+ 7
579
+
580
+
581
+
582
+ [Gloo] Rank 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
583
+ [Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
584
+ [Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
585
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
586
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
587
+ [Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
588
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
589
+ [Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
590
+ [Gloo] Rank [Gloo] Rank 7 is connected to 76 peer ranks. is connected to Expected number of connected peer ranks is : 77 peer ranks.
591
+ Expected number of connected peer ranks is : 7
592
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
593
+ > initialized tensor model parallel with size 1
594
+ > initialized pipeline model parallel with size 1
595
+ > setting random seeds to 1234 ...
596
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
597
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
598
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
599
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
600
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
601
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
602
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
603
+ [rank7]:[W921 08:34:44.940583273 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.
604
+ [rank6]:[W921 08:34:44.942483532 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.
605
+ [rank1]:[W921 08:34:44.942609558 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.
606
+ [rank4]:[W921 08:34:44.942644956 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.
607
+ [rank3]:[W921 08:34:44.942673657 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.
608
+ [rank5]:[W921 08:34:44.942692935 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.
609
+ [rank2]:[W921 08:34:44.942824371 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.
610
+ > compiling dataset index builder ...
611
+ make: Entering directory '/workspace/Megatron-MoE-Echo/megatron/core/datasets'
612
+ make: Nothing to be done for 'default'.
613
+ make: Leaving directory '/workspace/Megatron-MoE-Echo/megatron/core/datasets'
614
+ >>> done with dataset index builder. Compilation time: 0.033 seconds
615
+ WARNING: constraints for invoking optimized fused softmax kernel are not met. We default back to unfused kernel invocations.
616
+ > compiling and loading fused kernels ...
617
+ [rank0]:[W921 08:34:44.100960440 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.
618
+ >>> done with compiling and loading fused kernels. Compilation time: 3.972 seconds
619
+ time to initialize megatron (seconds): 7.387
620
+ [after megatron is initialized] datetime: 2026-09-21 08:34:50
621
+ building GPT model ...
622
+ /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.
623
+ warnings.warn(
624
+ /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.
625
+ warnings.warn(
626
+ /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.
627
+ warnings.warn(
628
+ /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.
629
+ warnings.warn(
630
+ /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.
631
+ warnings.warn(
632
+ /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.
633
+ warnings.warn(
634
+ /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.
635
+ warnings.warn(
636
+ /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.
637
+ warnings.warn(
638
+ > number of parameters on (tensor, pipeline) model parallel rank (0, 0): 1281951744
639
+ 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=False, 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)
640
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 7
641
+ Params for bucket 1 (103022592 elements, 103022592 padded size):
642
+ module.output_layer.weight
643
+ Params for bucket 2 (46542848 elements, 46542848 padded size):
644
+ module.decoder.layers.15.self_attention.linear_qkv.weight
645
+ module.decoder.layers.15.pre_mlp_layernorm.weight
646
+ module.decoder.layers.15.mlp.router.weight
647
+ module.decoder.layers.15.self_attention.linear_proj.weight
648
+ module.decoder.layers.14.mlp.router.weight
649
+ module.decoder.layers.14.self_attention.linear_qkv.weight
650
+ module.decoder.layers.14.self_attention.linear_qkv.layer_norm_weight
651
+ module.decoder.layers.14.self_attention.linear_proj.weight
652
+ module.decoder.layers.13.mlp.router.weight
653
+ module.decoder.layers.13.pre_mlp_layernorm.weight
654
+ module.decoder.final_layernorm.weight
655
+ module.decoder.layers.15.self_attention.linear_qkv.layer_norm_weight
656
+ module.decoder.layers.14.pre_mlp_layernorm.weight
657
+ module.decoder.layers.13.self_attention.linear_qkv.weight
658
+ Params for bucket 3 (50737152 elements, 50737152 padded size):
659
+ module.decoder.layers.12.self_attention.linear_qkv.weight
660
+ module.decoder.layers.13.self_attention.linear_proj.weight
661
+ module.decoder.layers.11.mlp.router.weight
662
+ module.decoder.layers.11.pre_mlp_layernorm.weight
663
+ module.decoder.layers.10.mlp.router.weight
664
+ module.decoder.layers.10.pre_mlp_layernorm.weight
665
+ module.decoder.layers.13.self_attention.linear_qkv.layer_norm_weight
666
+ module.decoder.layers.12.mlp.router.weight
667
+ module.decoder.layers.12.pre_mlp_layernorm.weight
668
+ module.decoder.layers.12.self_attention.linear_qkv.layer_norm_weight
669
+ module.decoder.layers.12.self_attention.linear_proj.weight
670
+ module.decoder.layers.11.self_attention.linear_qkv.weight
671
+ module.decoder.layers.11.self_attention.linear_qkv.layer_norm_weight
672
+ module.decoder.layers.11.self_attention.linear_proj.weight
673
+ module.decoder.layers.10.self_attention.linear_qkv.weight
674
+ Params for bucket 4 (50737152 elements, 50737152 padded size):
675
+ module.decoder.layers.9.pre_mlp_layernorm.weight
676
+ module.decoder.layers.10.self_attention.linear_proj.weight
677
+ module.decoder.layers.9.mlp.router.weight
678
+ module.decoder.layers.8.mlp.router.weight
679
+ module.decoder.layers.8.pre_mlp_layernorm.weight
680
+ module.decoder.layers.7.mlp.router.weight
681
+ module.decoder.layers.7.pre_mlp_layernorm.weight
682
+ module.decoder.layers.10.self_attention.linear_qkv.layer_norm_weight
683
+ module.decoder.layers.9.self_attention.linear_qkv.weight
684
+ module.decoder.layers.9.self_attention.linear_qkv.layer_norm_weight
685
+ module.decoder.layers.9.self_attention.linear_proj.weight
686
+ module.decoder.layers.8.self_attention.linear_qkv.weight
687
+ module.decoder.layers.8.self_attention.linear_qkv.layer_norm_weight
688
+ module.decoder.layers.8.self_attention.linear_proj.weight
689
+ module.decoder.layers.7.self_attention.linear_qkv.weight
690
+ Params for bucket 5 (50737152 elements, 50737152 padded size):
691
+ module.decoder.layers.7.self_attention.linear_qkv.layer_norm_weight
692
+ module.decoder.layers.7.self_attention.linear_proj.weight
693
+ module.decoder.layers.6.self_attention.linear_qkv.weight
694
+ module.decoder.layers.6.pre_mlp_layernorm.weight
695
+ module.decoder.layers.6.mlp.router.weight
696
+ module.decoder.layers.6.self_attention.linear_proj.weight
697
+ module.decoder.layers.5.mlp.router.weight
698
+ module.decoder.layers.5.pre_mlp_layernorm.weight
699
+ module.decoder.layers.4.mlp.router.weight
700
+ module.decoder.layers.4.pre_mlp_layernorm.weight
701
+ module.decoder.layers.4.self_attention.linear_qkv.weight
702
+ module.decoder.layers.6.self_attention.linear_qkv.layer_norm_weight
703
+ module.decoder.layers.5.self_attention.linear_qkv.weight
704
+ module.decoder.layers.5.self_attention.linear_qkv.layer_norm_weight
705
+ module.decoder.layers.5.self_attention.linear_proj.weight
706
+ Params for bucket 6 (50737152 elements, 50737152 padded size):
707
+ module.decoder.layers.3.mlp.router.weight
708
+ module.decoder.layers.4.self_attention.linear_proj.weight
709
+ module.decoder.layers.3.self_attention.linear_proj.weight
710
+ module.decoder.layers.2.mlp.router.weight
711
+ module.decoder.layers.2.self_attention.linear_proj.weight
712
+ module.decoder.layers.1.mlp.router.weight
713
+ module.decoder.layers.1.pre_mlp_layernorm.weight
714
+ module.decoder.layers.4.self_attention.linear_qkv.layer_norm_weight
715
+ module.decoder.layers.3.pre_mlp_layernorm.weight
716
+ module.decoder.layers.3.self_attention.linear_qkv.weight
717
+ module.decoder.layers.3.self_attention.linear_qkv.layer_norm_weight
718
+ module.decoder.layers.2.pre_mlp_layernorm.weight
719
+ module.decoder.layers.2.self_attention.linear_qkv.weight
720
+ module.decoder.layers.2.self_attention.linear_qkv.layer_norm_weight
721
+ module.decoder.layers.1.self_attention.linear_qkv.weight
722
+ Params for bucket 7 (124131328 elements, 124131328 padded size):
723
+ module.decoder.layers.0.self_attention.linear_qkv.weight
724
+ module.decoder.layers.1.self_attention.linear_proj.weight
725
+ module.decoder.layers.0.mlp.router.weight
726
+ module.decoder.layers.0.self_attention.linear_qkv.layer_norm_weight
727
+ module.embedding.word_embeddings.weight
728
+ module.decoder.layers.1.self_attention.linear_qkv.layer_norm_weight
729
+ module.decoder.layers.0.pre_mlp_layernorm.weight
730
+ module.decoder.layers.0.self_attention.linear_proj.weight
731
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 20
732
+ Params for bucket 1 (41943040 elements, 41943040 padded size):
733
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight4
734
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight6
735
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight2
736
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight0
737
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight7
738
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight5
739
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight3
740
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight7
741
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight5
742
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight3
743
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight1
744
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight6
745
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight4
746
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight2
747
+ Params for bucket 2 (41943040 elements, 41943040 padded size):
748
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight1
749
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight6
750
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight4
751
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight2
752
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight0
753
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight7
754
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight5
755
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight7
756
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight0
757
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight5
758
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight3
759
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight1
760
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight6
761
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight4
762
+ Params for bucket 3 (41943040 elements, 41943040 padded size):
763
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight3
764
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight1
765
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight6
766
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight4
767
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight2
768
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight0
769
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight7
770
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight2
771
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight7
772
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight0
773
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight5
774
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight3
775
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight1
776
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight6
777
+ Params for bucket 4 (41943040 elements, 41943040 padded size):
778
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight5
779
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight1
780
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight3
781
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight6
782
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight4
783
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight2
784
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight0
785
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight4
786
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight2
787
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight0
788
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight7
789
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight5
790
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight3
791
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight1
792
+ Params for bucket 5 (41943040 elements, 41943040 padded size):
793
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight7
794
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight5
795
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight3
796
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight1
797
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight7
798
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight5
799
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight6
800
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight4
801
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight2
802
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight0
803
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight6
804
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight4
805
+ Params for bucket 6 (41943040 elements, 41943040 padded size):
806
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight3
807
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight1
808
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight6
809
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight4
810
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight2
811
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight0
812
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight2
813
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight7
814
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight0
815
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight5
816
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight3
817
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight1
818
+ Params for bucket 7 (41943040 elements, 41943040 padded size):
819
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight7
820
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight3
821
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight5
822
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight1
823
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight6
824
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight4
825
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight2
826
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight6
827
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight4
828
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight2
829
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight0
830
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight7
831
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight5
832
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight3
833
+ Params for bucket 8 (41943040 elements, 41943040 padded size):
834
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight6
835
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight0
836
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight4
837
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight2
838
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight0
839
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight7
840
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight5
841
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight1
842
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight7
843
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight5
844
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight3
845
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight1
846
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight6
847
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight4
848
+ Params for bucket 9 (41943040 elements, 41943040 padded size):
849
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight3
850
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight1
851
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight7
852
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight5
853
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight3
854
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight1
855
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight6
856
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight0
857
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight2
858
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight6
859
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight4
860
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight2
861
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight0
862
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight7
863
+ Params for bucket 10 (41943040 elements, 41943040 padded size):
864
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight4
865
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight2
866
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight0
867
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight7
868
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight5
869
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight3
870
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight1
871
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight5
872
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight3
873
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight1
874
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight6
875
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight4
876
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight2
877
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight0
878
+ Params for bucket 11 (41943040 elements, 41943040 padded size):
879
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight4
880
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight6
881
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight2
882
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight0
883
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight6
884
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight4
885
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight7
886
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight3
887
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight5
888
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight1
889
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight7
890
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight5
891
+ Params for bucket 12 (41943040 elements, 41943040 padded size):
892
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight2
893
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight0
894
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight6
895
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight4
896
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight2
897
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight0
898
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight3
899
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight1
900
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight7
901
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight5
902
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight3
903
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight1
904
+ Params for bucket 13 (41943040 elements, 41943040 padded size):
905
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight6
906
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight4
907
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight2
908
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight0
909
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight7
910
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight5
911
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight3
912
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight7
913
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight5
914
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight3
915
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight1
916
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight6
917
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight4
918
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight2
919
+ Params for bucket 14 (41943040 elements, 41943040 padded size):
920
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight1
921
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight7
922
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight5
923
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight3
924
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight1
925
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight6
926
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight4
927
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight6
928
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight0
929
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight4
930
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight2
931
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight0
932
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight7
933
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight5
934
+ Params for bucket 15 (41943040 elements, 41943040 padded size):
935
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight2
936
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight0
937
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight6
938
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight4
939
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight2
940
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight0
941
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight7
942
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight3
943
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight1
944
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight7
945
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight5
946
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight3
947
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight1
948
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight6
949
+ Params for bucket 16 (41943040 elements, 41943040 padded size):
950
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight5
951
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight3
952
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight1
953
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight7
954
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight5
955
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight3
956
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight1
957
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight2
958
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight4
959
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight0
960
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight6
961
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight4
962
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight2
963
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight0
964
+ Params for bucket 17 (41943040 elements, 41943040 padded size):
965
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight7
966
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight3
967
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight5
968
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight1
969
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight6
970
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight4
971
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight6
972
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight4
973
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight2
974
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight0
975
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight7
976
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight5
977
+ Params for bucket 18 (41943040 elements, 41943040 padded size):
978
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight0
979
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight2
980
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight7
981
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight5
982
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight3
983
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight1
984
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight3
985
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight1
986
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight6
987
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight4
988
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight2
989
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight0
990
+ Params for bucket 19 (41943040 elements, 41943040 padded size):
991
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight7
992
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight3
993
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight5
994
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight1
995
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight6
996
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight4
997
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight2
998
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight4
999
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight6
1000
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight2
1001
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight0
1002
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight7
1003
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight5
1004
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight3
1005
+ Params for bucket 20 (8388608 elements, 8388608 padded size):
1006
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight1
1007
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight0
1008
+ 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 0x7527042c98b0>, config_logger_dir='')
1009
+ INFO:megatron.core.optimizer_param_scheduler:> learning rate decay style: cosine
1010
+ [after model, optimizer, and learning rate scheduler are built] datetime: 2026-09-21 08:34:50
1011
+ > building train, validation, and test datasets ...
1012
+ > datasets target sizes (minimum size):
1013
+ train: 96
1014
+ validation: 0
1015
+ test: 0
1016
+ INFO:megatron.core.datasets.blended_megatron_dataset_config:Let split_matrix = [(0, 1.0), None, None]
1017
+ > building train, validation, and test datasets for GPT ...
1018
+ INFO:megatron.core.datasets.blended_megatron_dataset_builder:Building GPTDataset splits with sizes=(96, 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 0x75270439ba40>, 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)
1019
+ INFO:megatron.core.datasets.indexed_dataset:Load the _IndexReader from /workspace/data/wikitext/wikitext103_gpt2_text_document.idx
1020
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence lengths
1021
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence pointers
1022
+ INFO:megatron.core.datasets.indexed_dataset: Extract the document indices
1023
+ INFO:megatron.core.datasets.indexed_dataset:> total number of sequences: 1165029
1024
+ INFO:megatron.core.datasets.indexed_dataset:> total number of documents: 1801350
1025
+ INFO:megatron.core.datasets.gpt_dataset:Load the GPTDataset train indices
1026
+ INFO:megatron.core.datasets.gpt_dataset: Load the document index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-document_index.npy
1027
+ INFO:megatron.core.datasets.gpt_dataset: Load the sample index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-sample_index.npy
1028
+ INFO:megatron.core.datasets.gpt_dataset: Load the shuffle index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-shuffle_index.npy
1029
+ INFO:megatron.core.datasets.gpt_dataset:> total number of samples: 29073
1030
+ > finished creating GPT datasets ...
1031
+ [after dataloaders are built] datetime: 2026-09-21 08:34:50
1032
+ done with setup ...
1033
+ (min, max) time across ranks (ms):
1034
+ model-and-optimizer-setup ......................: (286.08, 296.61)
1035
+ train/valid/test-data-iterators-setup ..........: (65.60, 74.13)
1036
+ training ...
1037
+ Overwriting rerun_state_machine.current_iteration from -1 to 0...
1038
+ [before the start of training step] datetime: 2026-09-21 08:34:50
1039
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1042
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1043
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1044
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1048
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1052
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1056
+ warnings.warn("The num_dispatched_tokens 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:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1062
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1063
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1066
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1067
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1070
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1071
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1072
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1073
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1074
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1075
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1076
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1077
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1078
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1079
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1080
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1081
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1082
+ warnings.warn("The num_dispatched_tokens 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:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1086
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1087
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1092
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1093
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1094
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1095
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1096
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1097
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1098
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1099
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1100
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1101
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1102
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1103
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1104
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1105
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1106
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1107
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1108
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1109
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1110
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1111
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1112
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1113
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1114
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1115
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1116
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1117
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1118
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1119
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1120
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1121
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1122
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1123
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1124
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1125
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1126
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1127
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1128
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1129
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1130
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1131
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1132
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1133
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1134
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1135
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1136
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1137
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1138
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1139
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1140
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1141
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1142
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1143
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1144
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1145
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1146
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1147
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1148
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1149
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1150
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1151
+ [2026-09-21 08:35:27] iteration 1/ 12 | consumed samples: 8 | elapsed time per iteration (ms): 36712.7 | throughput per GPU (TFLOP/s/GPU): 0.9 | learning rate: 2.000000E-05 | global batch size: 8 | lm loss: 1.118499E+01 | loss scale: 1.0 | grad norm: 35.477 | number of skipped iterations: 0 | number of nan iterations: 0 |
1152
+ Number of parameters in transformer block in billions: 6.71
1153
+ Number of parameters in embedding layers in billions: 0.21
1154
+ Total number of parameters in billions: 6.92
1155
+ Number of parameters in most loaded shard in billions: 6.9171
1156
+ Theoretical memory footprints: weight and optimizer=49474.72 MB
1157
+ [Rank 0] (after 1 iterations) memory (MB) | allocated: 19780.87158203125 | max allocated: 23491.14599609375 | reserved: 24842.0 | max reserved: 24842.0
1158
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1159
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1160
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1161
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1162
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1163
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1164
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1165
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1166
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1167
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1168
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1169
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1170
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1171
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1172
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1173
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1174
+ [2026-09-21 08:35:27] iteration 2/ 12 | consumed samples: 16 | elapsed time per iteration (ms): 343.6 | throughput per GPU (TFLOP/s/GPU): 93.8 | learning rate: 4.000000E-05 | global batch size: 8 | lm loss: 1.120955E+01 | loss scale: 1.0 | grad norm: 36.284 | number of skipped iterations: 0 | number of nan iterations: 0 |
1175
+ [2026-09-21 08:35:28] iteration 3/ 12 | consumed samples: 24 | elapsed time per iteration (ms): 547.2 | throughput per GPU (TFLOP/s/GPU): 58.9 | learning rate: 6.000000E-05 | global batch size: 8 | lm loss: 1.006655E+01 | loss scale: 1.0 | grad norm: 101.492 | number of skipped iterations: 0 | number of nan iterations: 0 |
1176
+ Capture CUDA graph for training!!!
1177
+ CUDA graph capture done!!!
1178
+ [2026-09-21 08:35:29] iteration 4/ 12 | consumed samples: 32 | elapsed time per iteration (ms): 931.1 | throughput per GPU (TFLOP/s/GPU): 34.6 | learning rate: 8.000000E-05 | global batch size: 8 | lm loss: 1.023759E+01 | loss scale: 1.0 | grad norm: 29.963 | number of skipped iterations: 0 | number of nan iterations: 0 |
1179
+ [2026-09-21 08:35:29] iteration 5/ 12 | consumed samples: 40 | elapsed time per iteration (ms): 113.6 | throughput per GPU (TFLOP/s/GPU): 283.6 | learning rate: 1.000000E-04 | global batch size: 8 | lm loss: 1.077593E+01 | loss scale: 1.0 | grad norm: 21.360 | number of skipped iterations: 0 | number of nan iterations: 0 |
1180
+ [2026-09-21 08:35:29] iteration 6/ 12 | consumed samples: 48 | elapsed time per iteration (ms): 117.5 | throughput per GPU (TFLOP/s/GPU): 274.2 | learning rate: 7.750000E-05 | global batch size: 8 | lm loss: 9.671508E+00 | loss scale: 1.0 | grad norm: 8.507 | number of skipped iterations: 0 | number of nan iterations: 0 |
1181
+ [2026-09-21 08:35:29] iteration 7/ 12 | consumed samples: 56 | elapsed time per iteration (ms): 111.1 | throughput per GPU (TFLOP/s/GPU): 290.0 | learning rate: 3.250000E-05 | global batch size: 8 | lm loss: 9.131283E+00 | loss scale: 1.0 | grad norm: 15.234 | number of skipped iterations: 0 | number of nan iterations: 0 |
1182
+ [2026-09-21 08:35:29] iteration 8/ 12 | consumed samples: 64 | elapsed time per iteration (ms): 112.1 | throughput per GPU (TFLOP/s/GPU): 287.5 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.840349E+00 | loss scale: 1.0 | grad norm: 6.898 | number of skipped iterations: 0 | number of nan iterations: 0 |
1183
+ [2026-09-21 08:35:29] iteration 9/ 12 | consumed samples: 72 | elapsed time per iteration (ms): 110.3 | throughput per GPU (TFLOP/s/GPU): 292.2 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.723488E+00 | loss scale: 1.0 | grad norm: 4.128 | number of skipped iterations: 0 | number of nan iterations: 0 |
1184
+ [2026-09-21 08:35:29] iteration 10/ 12 | consumed samples: 80 | elapsed time per iteration (ms): 110.4 | throughput per GPU (TFLOP/s/GPU): 291.9 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.659199E+00 | loss scale: 1.0 | grad norm: 3.607 | number of skipped iterations: 0 | number of nan iterations: 0 |
1185
+ [2026-09-21 08:35:29] iteration 11/ 12 | consumed samples: 88 | elapsed time per iteration (ms): 109.2 | throughput per GPU (TFLOP/s/GPU): 294.9 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.643869E+00 | loss scale: 1.0 | grad norm: 3.524 | number of skipped iterations: 0 | number of nan iterations: 0 |
1186
+ [2026-09-21 08:35:30] iteration 12/ 12 | consumed samples: 96 | elapsed time per iteration (ms): 109.6 | throughput per GPU (TFLOP/s/GPU): 293.9 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.591326E+00 | loss scale: 1.0 | grad norm: 3.418 | number of skipped iterations: 0 | number of nan iterations: 0 |
1187
+ [after training is done] datetime: 2026-09-21 08:35:30
1188
+ [rank0]:[W921 08:35:31.173653411 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())
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3/mem_samples.csv ADDED
@@ -0,0 +1,224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 0, 0
2
+ 1, 0
3
+ 2, 0
4
+ 3, 0
5
+ 4, 0
6
+ 5, 0
7
+ 6, 0
8
+ 7, 0
9
+ 0, 0
10
+ 1, 0
11
+ 2, 0
12
+ 3, 0
13
+ 4, 0
14
+ 5, 0
15
+ 6, 0
16
+ 7, 0
17
+ 0, 2
18
+ 1, 1
19
+ 2, 1
20
+ 3, 1
21
+ 4, 1
22
+ 5, 1
23
+ 6, 1
24
+ 7, 1
25
+ 0, 4
26
+ 1, 4
27
+ 2, 4
28
+ 3, 4
29
+ 4, 1
30
+ 5, 1
31
+ 6, 1
32
+ 7, 1
33
+ 0, 4
34
+ 1, 4
35
+ 2, 4
36
+ 3, 4
37
+ 4, 4
38
+ 5, 4
39
+ 6, 4
40
+ 7, 4
41
+ 0, 4
42
+ 1, 629
43
+ 2, 4
44
+ 3, 4
45
+ 4, 4
46
+ 5, 4
47
+ 6, 4
48
+ 7, 4
49
+ 0, 821
50
+ 1, 821
51
+ 2, 821
52
+ 3, 821
53
+ 4, 821
54
+ 5, 821
55
+ 6, 821
56
+ 7, 821
57
+ 0, 1803
58
+ 1, 1803
59
+ 2, 1803
60
+ 3, 1803
61
+ 4, 1803
62
+ 5, 1803
63
+ 6, 1803
64
+ 7, 1803
65
+ 0, 13147
66
+ 1, 13161
67
+ 2, 13123
68
+ 3, 13137
69
+ 4, 13139
70
+ 5, 13191
71
+ 6, 13185
72
+ 7, 13153
73
+ 0, 13775
74
+ 1, 13773
75
+ 2, 13811
76
+ 3, 13801
77
+ 4, 13783
78
+ 5, 13801
79
+ 6, 13813
80
+ 7, 13801
81
+ 0, 14831
82
+ 1, 14833
83
+ 2, 14833
84
+ 3, 14835
85
+ 4, 14829
86
+ 5, 14833
87
+ 6, 14835
88
+ 7, 14829
89
+ 0, 14915
90
+ 1, 14917
91
+ 2, 14917
92
+ 3, 14919
93
+ 4, 14913
94
+ 5, 14917
95
+ 6, 14919
96
+ 7, 14913
97
+ 0, 14915
98
+ 1, 14917
99
+ 2, 14917
100
+ 3, 14919
101
+ 4, 14913
102
+ 5, 14917
103
+ 6, 14919
104
+ 7, 14913
105
+ 0, 15561
106
+ 1, 15563
107
+ 2, 15563
108
+ 3, 15565
109
+ 4, 15559
110
+ 5, 15563
111
+ 6, 15565
112
+ 7, 15559
113
+ 0, 15711
114
+ 1, 15713
115
+ 2, 15713
116
+ 3, 15715
117
+ 4, 15709
118
+ 5, 15713
119
+ 6, 15715
120
+ 7, 15709
121
+ 0, 15851
122
+ 1, 15853
123
+ 2, 15853
124
+ 3, 15855
125
+ 4, 15849
126
+ 5, 15853
127
+ 6, 15855
128
+ 7, 15849
129
+ 0, 16349
130
+ 1, 16351
131
+ 2, 16351
132
+ 3, 16353
133
+ 4, 16347
134
+ 5, 16351
135
+ 6, 16351
136
+ 7, 16347
137
+ 0, 27465
138
+ 1, 27467
139
+ 2, 27477
140
+ 3, 27469
141
+ 4, 27399
142
+ 5, 27467
143
+ 6, 27469
144
+ 7, 27463
145
+ 0, 28963
146
+ 1, 28965
147
+ 2, 28975
148
+ 3, 28967
149
+ 4, 28961
150
+ 5, 28965
151
+ 6, 28967
152
+ 7, 28961
153
+ 0, 28963
154
+ 1, 28965
155
+ 2, 28975
156
+ 3, 28967
157
+ 4, 28961
158
+ 5, 28965
159
+ 6, 28967
160
+ 7, 28961
161
+ 0, 28963
162
+ 1, 28965
163
+ 2, 28975
164
+ 3, 28967
165
+ 4, 28961
166
+ 5, 28965
167
+ 6, 28967
168
+ 7, 28961
169
+ 0, 28963
170
+ 1, 28965
171
+ 2, 28975
172
+ 3, 28967
173
+ 4, 28961
174
+ 5, 28965
175
+ 6, 28967
176
+ 7, 28961
177
+ 0, 29395
178
+ 1, 29377
179
+ 2, 29389
180
+ 3, 29401
181
+ 4, 29435
182
+ 5, 29423
183
+ 6, 29439
184
+ 7, 29461
185
+ 0, 30349
186
+ 1, 30223
187
+ 2, 30233
188
+ 3, 30225
189
+ 4, 30603
190
+ 5, 30223
191
+ 6, 30609
192
+ 7, 30731
193
+ 0, 30915
194
+ 1, 30803
195
+ 2, 30815
196
+ 3, 30809
197
+ 4, 31175
198
+ 5, 30793
199
+ 6, 31195
200
+ 7, 31321
201
+ 0, 42895
202
+ 1, 43043
203
+ 2, 42911
204
+ 3, 42905
205
+ 4, 43035
206
+ 5, 42903
207
+ 6, 42905
208
+ 7, 43035
209
+ 0, 42745
210
+ 1, 42893
211
+ 2, 42761
212
+ 3, 42755
213
+ 4, 42885
214
+ 5, 42753
215
+ 6, 42755
216
+ 7, 42885
217
+ 0, 563
218
+ 1, 566
219
+ 2, 566
220
+ 3, 566
221
+ 4, 822
222
+ 5, 42107
223
+ 6, 1846
224
+ 7, 566
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3/run.env ADDED
@@ -0,0 +1,102 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ backend=echo
2
+ config=OLMOE_1B_7B_L16
3
+ family=tierB
4
+ source_model=OLMoE-1B-7B-0924
5
+ standard_num_layers=16
6
+ base_rung=
7
+ repo=/workspace/Megatron-MoE-Echo
8
+ branch=bracing-regions-through-kernels
9
+ commit=9a236793
10
+ num_layers=16
11
+ hidden=2048
12
+ ffn_hidden=1024
13
+ moe_ffn_hidden=1024
14
+ num_experts=64
15
+ topk=8
16
+ experts_per_rank=8
17
+ expert_params=6291456
18
+ heads=16
19
+ kv_groups=16
20
+ kv_channels=128
21
+ qk_layernorm=0
22
+ shared_ffn=0
23
+ shared_expert_overlap=0
24
+ moe_token_dispatcher=flex
25
+ mla=0
26
+ q_lora_rank=
27
+ kv_lora_rank=512
28
+ qk_head_dim=128
29
+ qk_pos_emb_head_dim=64
30
+ v_head_dim=128
31
+ dense_first_layers=0
32
+ moe_layers=16
33
+ rotary_base=10000
34
+ rotary_percent=1.0
35
+ rotary_scaling=
36
+ mscale=
37
+ mscale_all_dim=
38
+ qkv_bias=0
39
+ router_balancing=aux_loss
40
+ aux_loss_coeff=1e-2
41
+ router_pre_softmax=1
42
+ router_fusion=0
43
+ router_dtype=fp32
44
+ router_groups=
45
+ router_group_topk=
46
+ router_score_function=softmax
47
+ router_expert_bias=0
48
+ router_bias_update_rate=
49
+ router_topk_scaling_factor=
50
+ shared_expert_gate=0
51
+ apply_layernorm_1p=0
52
+ attention_output_gate=0
53
+ linear_attention_type=
54
+ linear_attention_freq=
55
+ linear_conv_kernel_dim=
56
+ linear_key_head_dim=
57
+ linear_value_head_dim=
58
+ linear_num_key_heads=
59
+ linear_num_value_heads=
60
+ mtp_num_layers=0
61
+ mtp_loss_scaling_factor=0.1
62
+ no_weight_decay_cond_type=
63
+ max_position_embeddings=4096
64
+ init_method_std=0.02
65
+ device_initiated_grouped_gemm=1
66
+ norm_epsilon=1e-5
67
+ seq_length=4096
68
+ micro_batch_size=1
69
+ global_batch_size=8
70
+ num_microbatches=1
71
+ assignments_per_rank=32768
72
+ mean_tokens_per_expert=4096
73
+ slot_policy=matched
74
+ slots_per_rank=8
75
+ capacity_mode=static
76
+ capacity=1.08984375
77
+ cuda_graph=1
78
+ probe_mode=0
79
+ async_grad_reduce=0
80
+ branch_role=timing
81
+ memory_dump_mode=none
82
+ memory_snapshot_path=
83
+ cuda_graph_warmup_steps=3
84
+ recompute=none
85
+ force_balance=0
86
+ train_iters=12
87
+ ep_size=8
88
+ echo_dispatch_overlap=1
89
+ echo_dispatch_recompute=0
90
+ ultraep_quota_min_tokens=
91
+ ultraep_weight_sync_join=
92
+ profile_enable=0
93
+ kernel_trace=0
94
+ fp8_mode=mxfp8
95
+ fp8_recipe=mxfp8
96
+ training_seed=1234
97
+ learning_rate=1e-4
98
+ min_learning_rate=1.0e-5
99
+ lr_warmup_iters=5
100
+ state_dump_enable=0
101
+ state_dump_dir=
102
+ moonep_assert_no_pending_reduce=0
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3/train.log ADDED
@@ -0,0 +1,1220 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==============================================================
2
+ OLMOE_1B_7B_L16 / echo (slots/rank=8 policy=matched)
3
+ E=64 K=8 H'=1024 m=8
4
+ L=16 seq=4096 gbs=8 (1 microbatches)
5
+ cap_mode=static cap=1.08984375 cg=1 recompute=none
6
+ memory_dump=none snapshots=/workspace/granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3/memory_snapshots/snapshot.pickle
7
+ repo=/workspace/Megatron-MoE-Echo @ 9a236793
8
+ ==============================================================
9
+ W0921 08:35:37.243000 686861 torch/distributed/run.py:766]
10
+ W0921 08:35:37.243000 686861 torch/distributed/run.py:766] *****************************************
11
+ W0921 08:35:37.243000 686861 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
+ W0921 08:35:37.243000 686861 torch/distributed/run.py:766] *****************************************
13
+ /workspace/Megatron-MoE-Echo/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
+ 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
16
+ WARNING: overriding default arguments for tokenizer_type:GPT2BPETokenizer with tokenizer_type:GPT2BPETokenizer
17
+ Number of virtual stages per pipeline stage: None
18
+ accumulate and all-reduce gradients in fp32 for bfloat16 data type.
19
+ using torch.bfloat16 for parameters ...
20
+ /workspace/Megatron-MoE-Echo/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
+ ------------------------ arguments ------------------------
23
+ account_for_embedding_in_pipeline_split ......... False
24
+ account_for_loss_in_pipeline_split .............. False
25
+ accumulate_allreduce_grads_in_fp32 .............. True
26
+ adam_beta1 ...................................... 0.9
27
+ adam_beta2 ...................................... 0.999
28
+ adam_eps ........................................ 1e-08
29
+ add_bias_linear ................................. False
30
+ add_position_embedding .......................... False
31
+ add_qkv_bias .................................... False
32
+ adlr_autoresume ................................. False
33
+ adlr_autoresume_interval ........................ 1000
34
+ align_grad_reduce ............................... True
35
+ align_param_gather .............................. False
36
+ app_tag_run_name ................................ None
37
+ app_tag_run_version ............................. 0.0.0
38
+ apply_layernorm_1p .............................. False
39
+ apply_query_key_layer_scaling ................... False
40
+ apply_residual_connection_post_layernorm ........ False
41
+ apply_rope_fusion ............................... True
42
+ async_save ...................................... None
43
+ async_tensor_model_parallel_allreduce ........... True
44
+ attention_backend ............................... AttnBackend.auto
45
+ attention_dropout ............................... 0.0
46
+ attention_softmax_in_fp32 ....................... False
47
+ auto_detect_ckpt_format ......................... False
48
+ barrier_with_L1_time ............................ True
49
+ bert_binary_head ................................ True
50
+ bert_embedder_type .............................. megatron
51
+ bert_load ....................................... None
52
+ bf16 ............................................ True
53
+ bias_dropout_fusion ............................. True
54
+ bias_gelu_fusion ................................ False
55
+ bias_swiglu_fusion .............................. True
56
+ biencoder_projection_dim ........................ 0
57
+ biencoder_shared_query_context_model ............ False
58
+ block_data_path ................................. None
59
+ calc_ft_timeouts ................................ False
60
+ calculate_per_token_loss ........................ False
61
+ check_for_large_grads ........................... False
62
+ check_for_nan_in_loss_and_grad .................. False
63
+ check_for_spiky_loss ............................ False
64
+ check_weight_hash_across_dp_replicas_interval ... None
65
+ ckpt_assume_constant_structure .................. False
66
+ ckpt_convert_format ............................. None
67
+ ckpt_convert_save ............................... None
68
+ ckpt_convert_update_legacy_dist_opt_format ...... False
69
+ ckpt_format ..................................... torch_dist
70
+ ckpt_fully_parallel_load ........................ False
71
+ ckpt_fully_parallel_save ........................ True
72
+ ckpt_fully_parallel_save_deprecated ............. False
73
+ ckpt_step ....................................... None
74
+ classes_fraction ................................ 1.0
75
+ clip_grad ....................................... 1.0
76
+ clone_scatter_output_in_embedding ............... True
77
+ config_logger_dir ...............................
78
+ consumed_train_samples .......................... 0
79
+ consumed_valid_samples .......................... 0
80
+ context_parallel_size ........................... 1
81
+ cp_comm_type .................................... ['p2p']
82
+ create_attention_mask_in_dataloader ............. True
83
+ cross_entropy_fusion_impl ....................... native
84
+ cross_entropy_loss_fusion ....................... False
85
+ cuda_graph_scope ................................ full_iteration
86
+ cuda_graph_warmup_steps ......................... 3
87
+ data_args_path .................................. None
88
+ data_cache_path ................................. None
89
+ data_parallel_random_init ....................... False
90
+ data_parallel_sharding_strategy ................. no_shard
91
+ data_parallel_size .............................. 8
92
+ data_path ....................................... ['/workspace/data/wikitext/wikitext103_gpt2_text_document']
93
+ data_per_class_fraction ......................... 1.0
94
+ data_sharding ................................... True
95
+ dataloader_type ................................. single
96
+ ddp_average_in_collective ....................... False
97
+ ddp_bucket_size ................................. None
98
+ ddp_num_buckets ................................. None
99
+ ddp_pad_buckets_for_high_nccl_busbw ............. False
100
+ decoder_first_pipeline_num_layers ............... None
101
+ decoder_last_pipeline_num_layers ................ None
102
+ decoder_num_layers .............................. None
103
+ decoder_seq_length .............................. None
104
+ decoupled_lr .................................... None
105
+ decoupled_min_lr ................................ None
106
+ decrease_batch_size_if_needed ................... False
107
+ defer_embedding_wgrad_compute ................... False
108
+ delay_wgrad_compute ............................. False
109
+ deprecated_use_mcore_models ..................... True
110
+ deterministic_mode .............................. False
111
+ dino_bottleneck_size ............................ 256
112
+ dino_freeze_last_layer .......................... 1
113
+ dino_head_hidden_size ........................... 2048
114
+ dino_local_crops_number ......................... 10
115
+ dino_local_img_size ............................. 96
116
+ dino_norm_last_layer ............................ False
117
+ dino_teacher_temp ............................... 0.07
118
+ dino_warmup_teacher_temp ........................ 0.04
119
+ dino_warmup_teacher_temp_epochs ................. 30
120
+ disable_bf16_reduced_precision_matmul ........... False
121
+ disable_mamba_mem_eff_path ...................... False
122
+ disable_straggler_on_startup .................... False
123
+ dist_ckpt_format_deprecated ..................... None
124
+ dist_ckpt_strictness ............................ assume_ok_unexpected
125
+ distribute_saved_activations .................... False
126
+ distributed_backend ............................. nccl
127
+ distributed_timeout_minutes ..................... 10
128
+ embedding_init_method_std ....................... None
129
+ embedding_path .................................. None
130
+ empty_unused_memory_level ....................... 0
131
+ enable_cuda_graph ............................... True
132
+ enable_experimental ............................. False
133
+ enable_ft_package ............................... False
134
+ enable_gloo_process_groups ...................... True
135
+ enable_msc ...................................... True
136
+ enable_one_logger ............................... True
137
+ encoder_num_layers .............................. 16
138
+ encoder_seq_length .............................. 4096
139
+ end_weight_decay ................................ 0.1
140
+ eod_mask_loss ................................... False
141
+ error_injection_rate ............................ 0
142
+ error_injection_type ............................ transient_error
143
+ eval_interval ................................... 1000000
144
+ eval_iters ...................................... 0
145
+ evidence_data_path .............................. None
146
+ exit_duration_in_mins ........................... None
147
+ exit_interval ................................... None
148
+ exit_on_missing_checkpoint ...................... False
149
+ exit_signal_handler ............................. False
150
+ exp_avg_dtype ................................... torch.float32
151
+ exp_avg_sq_dtype ................................ torch.float32
152
+ expert_model_parallel_size ...................... 8
153
+ expert_tensor_parallel_size ..................... 1
154
+ export_force_local_attention .................... False
155
+ export_kd_cfg ................................... None
156
+ export_kd_teacher_ckpt_format ................... None
157
+ export_kd_teacher_load .......................... None
158
+ export_kv_cache_quant ........................... False
159
+ export_legacy_megatron .......................... False
160
+ export_model_type ............................... GPTModel
161
+ export_moe_apply_probs_on_input ................. False
162
+ export_qk_l2_norm ............................... False
163
+ export_quant_cfg ................................ None
164
+ export_real_quant_cfg ........................... None
165
+ export_te_mcore_model ........................... False
166
+ external_cuda_graph ............................. False
167
+ ffn_hidden_size ................................. 1024
168
+ finetune ........................................ False
169
+ finetune_data_split ............................. train
170
+ finetune_hf_dataset ............................. None
171
+ first_last_layers_bf16 .......................... False
172
+ flash_decode .................................... False
173
+ fp16 ............................................ False
174
+ fp16_lm_cross_entropy ........................... False
175
+ fp32_residual_connection ........................ False
176
+ fp8 ............................................. e4m3
177
+ fp8_amax_compute_algo ........................... most_recent
178
+ fp8_amax_history_len ............................ 1
179
+ fp8_interval .................................... 1
180
+ fp8_margin ...................................... 0
181
+ fp8_param_gather ................................ False
182
+ fp8_recipe ...................................... mxfp8
183
+ fp8_wgrad ....................................... True
184
+ fsdp_double_buffer .............................. False
185
+ global_batch_size ............................... 8
186
+ grad_reduce_in_bf16 ............................. False
187
+ gradient_accumulation_fusion .................... True
188
+ gradient_reduce_div_fusion ...................... True
189
+ group_query_attention ........................... True
190
+ head_lr_mult .................................... 1.0
191
+ heterogeneous_layers_config_encoded_json ........ None
192
+ heterogeneous_layers_config_path ................ None
193
+ hidden_dropout .................................. 0.0
194
+ hidden_size ..................................... 2048
195
+ hierarchical_context_parallel_sizes ............. None
196
+ high_priority_stream_groups ..................... []
197
+ hybrid_attention_ratio .......................... 0.0
198
+ hybrid_mlp_ratio ................................ 0.0
199
+ hybrid_override_pattern ......................... None
200
+ hysteresis ...................................... 2
201
+ ict_head_size ................................... None
202
+ ict_load ........................................ None
203
+ img_h ........................................... 224
204
+ img_w ........................................... 224
205
+ indexer_batch_size .............................. 128
206
+ indexer_log_interval ............................ 1000
207
+ inference_batch_times_seqlen_threshold .......... -1
208
+ inference_dynamic_batching ...................... False
209
+ inference_dynamic_batching_buffer_guaranteed_fraction 0.2
210
+ inference_dynamic_batching_buffer_overflow_factor None
211
+ inference_dynamic_batching_buffer_size_gb ....... 40.0
212
+ inference_dynamic_batching_chunk_size ........... 256
213
+ inference_dynamic_batching_max_requests_override None
214
+ inference_dynamic_batching_max_tokens_override .. None
215
+ inference_dynamic_batching_num_cuda_graphs ...... 16
216
+ inference_max_batch_size ........................ 8
217
+ inference_max_seq_length ........................ 2560
218
+ inference_rng_tracker ........................... False
219
+ init_method_std ................................. 0.02
220
+ init_method_xavier_uniform ...................... False
221
+ init_model_with_meta_device ..................... False
222
+ initial_loss_scale .............................. 4294967296
223
+ inprocess_active_world_size ..................... 8
224
+ inprocess_barrier_timeout ....................... 120
225
+ inprocess_completion_timeout .................... 120
226
+ inprocess_empty_cuda_cache ...................... False
227
+ inprocess_granularity ........................... node
228
+ inprocess_hard_timeout .......................... 90
229
+ inprocess_heartbeat_interval .................... 30
230
+ inprocess_heartbeat_timeout ..................... 60
231
+ inprocess_last_call_wait ........................ 1
232
+ inprocess_max_iterations ........................ None
233
+ inprocess_monitor_process_interval .............. 1.0
234
+ inprocess_monitor_thread_interval ............... 1.0
235
+ inprocess_progress_watchdog_interval ............ 1.0
236
+ inprocess_restart ............................... False
237
+ inprocess_soft_timeout .......................... 60
238
+ inprocess_termination_grace_time ................ 1
239
+ is_hybrid_model ................................. False
240
+ iter_per_epoch .................................. 1250
241
+ iterations_to_skip .............................. []
242
+ keep_fp8_transpose_cache_when_using_custom_fsdp . False
243
+ kitchen_config_file ............................. None
244
+ kitchen_recipe_number ........................... None
245
+ kv_channels ..................................... 128
246
+ kv_lora_rank .................................... 32
247
+ lazy_mpu_init ................................... None
248
+ load ............................................ None
249
+ load_main_params_from_ckpt ...................... None
250
+ load_model_opt_format ........................... False
251
+ local_rank ...................................... 0
252
+ log_energy ...................................... False
253
+ log_interval .................................... 1
254
+ log_loss_scale_to_tensorboard ................... True
255
+ log_memory_to_tensorboard ....................... False
256
+ log_num_zeros_in_grad ........................... False
257
+ log_params_norm ................................. False
258
+ log_progress .................................... False
259
+ log_straggler ................................... False
260
+ log_throughput .................................. True
261
+ log_timers_to_tensorboard ....................... False
262
+ log_validation_ppl_to_tensorboard ............... False
263
+ log_world_size_to_tensorboard ................... False
264
+ logging_level ................................... None
265
+ loss_scale ...................................... None
266
+ loss_scale_window ............................... 1000
267
+ lr .............................................. 0.0001
268
+ lr_decay_iters .................................. 8
269
+ lr_decay_samples ................................ None
270
+ lr_decay_style .................................. cosine
271
+ lr_warmup_fraction .............................. None
272
+ lr_warmup_init .................................. 0.0
273
+ lr_warmup_iters ................................. 5
274
+ lr_warmup_samples ............................... 0
275
+ lr_wsd_decay_iters .............................. None
276
+ lr_wsd_decay_samples ............................ None
277
+ lr_wsd_decay_style .............................. exponential
278
+ main_grads_dtype ................................ torch.float32
279
+ main_params_dtype ............................... torch.float32
280
+ make_vocab_size_divisible_by .................... 128
281
+ mamba_head_dim .................................. 64
282
+ mamba_num_groups ................................ 8
283
+ mamba_num_heads ................................. None
284
+ mamba_state_dim ................................. 128
285
+ manual_gc ....................................... False
286
+ manual_gc_eval .................................. True
287
+ manual_gc_interval .............................. 0
288
+ mask_factor ..................................... 1.0
289
+ mask_prob ....................................... 0.15
290
+ mask_type ....................................... random
291
+ masked_softmax_fusion ........................... False
292
+ max_position_embeddings ......................... 4096
293
+ max_tokens_to_oom ............................... 12000
294
+ memory_snapshot_path ............................ snapshot.pickle
295
+ merge_file ...................................... /workspace/data/wikitext/gpt2-merges.txt
296
+ micro_batch_size ................................ 1
297
+ microbatch_group_size_per_vp_stage .............. None
298
+ mid_level_dataset_surplus ....................... 0.005
299
+ min_loss_scale .................................. 1.0
300
+ min_lr .......................................... 1e-05
301
+ mlp_chunks_for_prefill .......................... 1
302
+ mmap_bin_files .................................. True
303
+ mock_data ....................................... False
304
+ moe_apply_probs_on_input ........................ False
305
+ moe_aux_loss_coeff .............................. 0.01
306
+ moe_deepep_num_sms .............................. 20
307
+ moe_echo_enable_random_offloading ............... False
308
+ moe_echo_expert_dispatch_overlap ................ True
309
+ moe_echo_expert_dispatcher_type ................. hybridep
310
+ moe_echo_recompute_expert_dispatch .............. False
311
+ moe_enable_deepep ............................... False
312
+ moe_enable_echo ................................. True
313
+ moe_enable_hybridep ............................. True
314
+ moe_expert_capacity_factor ...................... None
315
+ moe_extended_tp ................................. False
316
+ moe_ffn_hidden_size ............................. 1024
317
+ moe_grouped_gemm ................................ True
318
+ moe_hybridep_num_sms ............................ 16
319
+ moe_input_jitter_eps ............................ None
320
+ moe_layer_freq .................................. 1
321
+ moe_layer_recompute ............................. False
322
+ moe_num_echo_experts ............................ 64
323
+ moe_pad_expert_input_to_capacity ................ False
324
+ moe_per_layer_logging ........................... False
325
+ moe_permute_fusion .............................. False
326
+ moe_received_token_capacity ..................... 1.08984375
327
+ moe_router_bias_update_rate ..................... 0.001
328
+ moe_router_dtype ................................ fp32
329
+ moe_router_enable_expert_bias ................... False
330
+ moe_router_force_load_balancing ................. False
331
+ moe_router_group_topk ........................... None
332
+ moe_router_load_balancing_type .................. aux_loss
333
+ moe_router_num_groups ........................... None
334
+ moe_router_padding_for_fp8 ...................... False
335
+ moe_router_pre_softmax .......................... True
336
+ moe_router_score_function ....................... softmax
337
+ moe_router_topk ................................. 8
338
+ moe_router_topk_scaling_factor .................. None
339
+ moe_shared_expert_intermediate_size ............. None
340
+ moe_shared_expert_overlap ....................... False
341
+ moe_token_dispatcher_type ....................... flex
342
+ moe_token_drop_policy ........................... probs
343
+ moe_upcycling_granularity ....................... 1
344
+ moe_use_device_initiated_grouped_gemm ........... True
345
+ moe_use_legacy_grouped_gemm ..................... False
346
+ moe_use_upcycling ............................... False
347
+ moe_z_loss_coeff ................................ None
348
+ mrope_section ................................... None
349
+ mscale .......................................... 1.0
350
+ mscale_all_dim .................................. 0.0
351
+ mtp_loss_scaling_factor ......................... 0.1
352
+ mtp_num_layers .................................. None
353
+ multi_latent_attention .......................... False
354
+ nccl_all_reduce_for_prefill ..................... False
355
+ nccl_communicator_config_path ................... None
356
+ nccl_ub ......................................... False
357
+ no_load_optim ................................... None
358
+ no_load_rng ..................................... None
359
+ no_persist_layer_norm ........................... False
360
+ no_rope_freq .................................... None
361
+ no_save_optim ................................... None
362
+ no_save_rng ..................................... None
363
+ non_persistent_ckpt_type ........................ None
364
+ non_persistent_global_ckpt_dir .................. None
365
+ non_persistent_local_ckpt_algo .................. fully_parallel
366
+ non_persistent_local_ckpt_dir ................... None
367
+ non_persistent_save_interval .................... None
368
+ norm_epsilon .................................... 1e-05
369
+ normalization ................................... RMSNorm
370
+ num_attention_heads ............................. 16
371
+ num_channels .................................... 3
372
+ num_classes ..................................... 1000
373
+ num_dataset_builder_threads ..................... 1
374
+ num_distributed_optimizer_instances ............. 1
375
+ num_experts ..................................... 64
376
+ num_layers ...................................... 16
377
+ num_layers_at_end_in_bf16 ....................... 1
378
+ num_layers_at_start_in_bf16 ..................... 1
379
+ num_layers_per_virtual_pipeline_stage ........... None
380
+ num_query_groups ................................ 16
381
+ num_virtual_stages_per_pipeline_rank ............ None
382
+ num_workers ..................................... 2
383
+ object_storage_cache_path ....................... None
384
+ one_logger_async ................................ False
385
+ one_logger_project .............................. megatron-lm
386
+ one_logger_run_name ............................. None
387
+ onnx_safe ....................................... None
388
+ openai_gelu ..................................... False
389
+ optimizer ....................................... adam
390
+ optimizer_cpu_offload ........................... False
391
+ optimizer_offload_fraction ...................... 1.0
392
+ output_bert_embeddings .......................... False
393
+ overlap_cpu_optimizer_d2h_h2d ................... False
394
+ overlap_grad_reduce ............................. True
395
+ overlap_moe_expert_parallel_comm ................ False
396
+ overlap_p2p_comm ................................ False
397
+ overlap_p2p_comm_warmup_flush ................... False
398
+ overlap_param_gather ............................ True
399
+ overlap_param_gather_with_optimizer_step ........ False
400
+ override_opt_param_scheduler .................... False
401
+ params_dtype .................................... torch.bfloat16
402
+ patch_dim ....................................... 16
403
+ per_split_data_args_path ........................ None
404
+ perform_initialization .......................... True
405
+ pin_cpu_grads ................................... True
406
+ pin_cpu_params .................................. True
407
+ pipeline_model_parallel_comm_backend ............ None
408
+ pipeline_model_parallel_layout .................. None
409
+ pipeline_model_parallel_size .................... 1
410
+ position_embedding_type ......................... rope
411
+ pretrained_checkpoint ........................... None
412
+ profile ......................................... False
413
+ profile_ranks ................................... [0]
414
+ profile_step_end ................................ 12
415
+ profile_step_start .............................. 10
416
+ q_lora_rank ..................................... None
417
+ qk_head_dim ..................................... 128
418
+ qk_l2_norm ...................................... False
419
+ qk_layernorm .................................... False
420
+ qk_pos_emb_head_dim ............................. 64
421
+ query_in_block_prob ............................. 0.1
422
+ rampup_batch_size ............................... None
423
+ rank ............................................ 0
424
+ recompute_granularity ........................... None
425
+ recompute_method ................................ None
426
+ recompute_modules ............................... None
427
+ recompute_num_layers ............................ None
428
+ record_memory_history ........................... False
429
+ relative_attention_max_distance ................. 128
430
+ relative_attention_num_buckets .................. 32
431
+ replication ..................................... False
432
+ replication_factor .............................. 2
433
+ replication_jump ................................ None
434
+ rerun_mode ...................................... validate_results
435
+ reset_attention_mask ............................ False
436
+ reset_position_ids .............................. False
437
+ result_rejected_tracker_filename ................ None
438
+ retriever_report_topk_accuracies ................ []
439
+ retriever_score_scaling ......................... False
440
+ retriever_seq_length ............................ 256
441
+ retro_add_retriever ............................. False
442
+ retro_attention_gate ............................ 1
443
+ retro_cyclic_train_iters ........................ None
444
+ retro_encoder_attention_dropout ................. 0.1
445
+ retro_encoder_hidden_dropout .................... 0.1
446
+ retro_encoder_layers ............................ 2
447
+ retro_num_neighbors ............................. 2
448
+ retro_num_retrieved_chunks ...................... 2
449
+ retro_project_dir ............................... None
450
+ retro_verify_neighbor_count ..................... True
451
+ reuse_grad_buf_for_mxfp8_param_ag ............... False
452
+ rope_scaling_factor ............................. 8.0
453
+ rope_type ....................................... None
454
+ rotary_base ..................................... 10000
455
+ rotary_interleaved .............................. False
456
+ rotary_percent .................................. 1.0
457
+ rotary_scaling_factor ........................... 1.0
458
+ rotary_seq_len_interpolation_factor ............. None
459
+ run_workload_inspector_server ................... False
460
+ sample_rate ..................................... 1.0
461
+ save ............................................ None
462
+ save_interval ................................... None
463
+ save_retain_interval ............................ None
464
+ scatter_gather_tensors_in_pipeline .............. True
465
+ seed ............................................ 1234
466
+ seq_length ...................................... 4096
467
+ sequence_parallel ............................... False
468
+ sft ............................................. False
469
+ sft_tokenizer_prompt_format ..................... nemotron-h-aligned
470
+ sgd_momentum .................................... 0.9
471
+ short_seq_prob .................................. 0.1
472
+ skip_train ...................................... False
473
+ skipped_train_samples ........................... 0
474
+ spec ............................................ None
475
+ split ........................................... 100,0,0
476
+ squared_relu .................................... False
477
+ start_weight_decay .............................. 0.1
478
+ straggler_ctrlr_port ............................ 65535
479
+ straggler_minmax_count .......................... 1
480
+ suggested_communication_unit_size ............... None
481
+ swiglu .......................................... True
482
+ swin_backbone_type .............................. tiny
483
+ symmetric_ar_type ............................... None
484
+ te_rng_tracker .................................. True
485
+ tensor_model_parallel_size ...................... 1
486
+ tensorboard_dir ................................. None
487
+ tensorboard_log_interval ........................ 1
488
+ tensorboard_queue_size .......................... 1000
489
+ test_data_path .................................. None
490
+ test_mode ....................................... False
491
+ tiktoken_num_special_tokens ..................... 1000
492
+ tiktoken_pattern ................................ None
493
+ tiktoken_special_tokens ......................... None
494
+ timing_log_level ................................ 0
495
+ timing_log_option ............................... minmax
496
+ titles_data_path ................................ None
497
+ tokenizer_model ................................. None
498
+ tokenizer_type .................................. GPT2BPETokenizer
499
+ torch_fsdp2_reshard_after_forward ............... True
500
+ tp_comm_bootstrap_backend ....................... nccl
501
+ tp_comm_bulk_dgrad .............................. True
502
+ tp_comm_bulk_wgrad .............................. True
503
+ tp_comm_overlap ................................. False
504
+ tp_comm_overlap_ag .............................. True
505
+ tp_comm_overlap_cfg ............................. None
506
+ tp_comm_overlap_rs .............................. True
507
+ tp_comm_overlap_rs_dgrad ........................ False
508
+ tp_comm_split_ag ................................ True
509
+ tp_comm_split_rs ................................ True
510
+ train_data_path ................................. None
511
+ train_iters ..................................... 12
512
+ train_samples ................................... None
513
+ train_sync_interval ............................. None
514
+ transformer_impl ................................ transformer_engine
515
+ transformer_pipeline_model_parallel_size ........ 1
516
+ untie_embeddings_and_output_weights ............. True
517
+ use_checkpoint_args ............................. False
518
+ use_checkpoint_opt_param_scheduler .............. False
519
+ use_cpu_initialization .......................... None
520
+ use_custom_fsdp ................................. False
521
+ use_dist_ckpt ................................... True
522
+ use_dist_ckpt_deprecated ........................ False
523
+ use_distributed_optimizer ....................... True
524
+ use_flash_attn .................................. False
525
+ use_legacy_models ............................... False
526
+ use_mp_args_from_checkpoint_args ................ False
527
+ use_one_sent_docs ............................... False
528
+ use_persistent_ckpt_worker ...................... False
529
+ use_precision_aware_optimizer ................... False
530
+ use_pytorch_profiler ............................ False
531
+ use_ring_exchange_p2p ........................... False
532
+ use_rope_scaling ................................ False
533
+ use_rotary_position_embeddings .................. False
534
+ use_sharp ....................................... False
535
+ use_tokenizer_model_from_checkpoint_args ........ True
536
+ use_torch_fsdp2 ................................. False
537
+ use_torch_optimizer_for_cpu_offload ............. False
538
+ use_tp_pp_dp_mapping ............................ False
539
+ v_head_dim ...................................... 128
540
+ valid_data_path ................................. None
541
+ variable_seq_lengths ............................ False
542
+ virtual_pipeline_model_parallel_size ............ None
543
+ vision_backbone_type ............................ vit
544
+ vision_pretraining .............................. False
545
+ vision_pretraining_type ......................... classify
546
+ vocab_extra_ids ................................. 0
547
+ vocab_file ...................................... /workspace/data/wikitext/gpt2-vocab.json
548
+ vocab_size ...................................... None
549
+ wandb_exp_name ..................................
550
+ wandb_project ...................................
551
+ wandb_save_dir ..................................
552
+ weight_decay .................................... 0.1
553
+ weight_decay_incr_style ......................... constant
554
+ wgrad_deferral_limit ............................ 0
555
+ world_size ...................................... 8
556
+ yaml_cfg ........................................ None
557
+ -------------------- end of arguments ---------------------
558
+ INFO:megatron.core.num_microbatches_calculator:setting number of microbatches to constant 1
559
+ > building GPT2BPETokenizer tokenizer ...
560
+ > padded vocab (size: 50257) with 47 dummy tokens (new size: 50304)
561
+ WARNING:megatron.core.rerun_state_machine:RerunStateMachine initialized in mode RerunMode.VALIDATE_RESULTS
562
+ > initializing torch distributed ...
563
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
564
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
565
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
566
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
567
+ 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
568
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
569
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
570
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
571
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
572
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
573
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
574
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
575
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
576
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
577
+ [Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
578
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
579
+ [Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
580
+ [Gloo] Rank 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
581
+ [Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
582
+ [Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
583
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
584
+ [Gloo] Rank [Gloo] Rank 3 is connected to 17 is connected to peer ranks. 7Expected number of connected peer ranks is : peer ranks. 7Expected number of connected peer ranks is :
585
+ 7
586
+ [Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
587
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
588
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
589
+ [Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
590
+ [Gloo] Rank 7 is connected to [Gloo] Rank 7 peer ranks. 6Expected number of connected peer ranks is : is connected to 77 peer ranks.
591
+ Expected number of connected peer ranks is : 7
592
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
593
+ > initialized tensor model parallel with size 1
594
+ > initialized pipeline model parallel with size 1
595
+ > setting random seeds to 1234 ...
596
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
597
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
598
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
599
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
600
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
601
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
602
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
603
+ [rank7]:[W921 08:35:53.885288162 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.
604
+ [rank4]:[W921 08:35:53.885357077 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.
605
+ [rank3]:[W921 08:35:53.885365610 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.
606
+ [rank2]:[W921 08:35:53.885374430 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.
607
+ [rank5]:[W921 08:35:53.885401659 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.
608
+ [rank1]:[W921 08:35:53.885479222 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.
609
+ [rank6]:[W921 08:35:53.885582764 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.
610
+ > compiling dataset index builder ...
611
+ make: Entering directory '/workspace/Megatron-MoE-Echo/megatron/core/datasets'
612
+ make: Nothing to be done for 'default'.
613
+ make: Leaving directory '/workspace/Megatron-MoE-Echo/megatron/core/datasets'
614
+ >>> done with dataset index builder. Compilation time: 0.035 seconds
615
+ WARNING: constraints for invoking optimized fused softmax kernel are not met. We default back to unfused kernel invocations.
616
+ > compiling and loading fused kernels ...
617
+ [rank0]:[W921 08:35:54.073849039 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.
618
+ >>> done with compiling and loading fused kernels. Compilation time: 3.908 seconds
619
+ time to initialize megatron (seconds): 7.328
620
+ [after megatron is initialized] datetime: 2026-09-21 08:36:00
621
+ building GPT model ...
622
+ /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.
623
+ warnings.warn(
624
+ /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.
625
+ warnings.warn(
626
+ /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.
627
+ warnings.warn(
628
+ /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.
629
+ warnings.warn(
630
+ /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.
631
+ warnings.warn(
632
+ /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.
633
+ warnings.warn(
634
+ /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.
635
+ warnings.warn(
636
+ /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.
637
+ warnings.warn(
638
+ > number of parameters on (tensor, pipeline) model parallel rank (0, 0): 1281951744
639
+ 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=False, 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)
640
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 7
641
+ Params for bucket 1 (103022592 elements, 103022592 padded size):
642
+ module.output_layer.weight
643
+ Params for bucket 2 (46542848 elements, 46542848 padded size):
644
+ module.decoder.layers.15.pre_mlp_layernorm.weight
645
+ module.decoder.layers.15.mlp.router.weight
646
+ module.decoder.layers.15.self_attention.linear_qkv.weight
647
+ module.decoder.layers.15.self_attention.linear_proj.weight
648
+ module.decoder.layers.14.pre_mlp_layernorm.weight
649
+ module.decoder.layers.14.self_attention.linear_qkv.weight
650
+ module.decoder.layers.14.self_attention.linear_qkv.layer_norm_weight
651
+ module.decoder.layers.14.self_attention.linear_proj.weight
652
+ module.decoder.layers.13.mlp.router.weight
653
+ module.decoder.layers.13.pre_mlp_layernorm.weight
654
+ module.decoder.final_layernorm.weight
655
+ module.decoder.layers.15.self_attention.linear_qkv.layer_norm_weight
656
+ module.decoder.layers.14.mlp.router.weight
657
+ module.decoder.layers.13.self_attention.linear_qkv.weight
658
+ Params for bucket 3 (50737152 elements, 50737152 padded size):
659
+ module.decoder.layers.12.pre_mlp_layernorm.weight
660
+ module.decoder.layers.12.mlp.router.weight
661
+ module.decoder.layers.13.self_attention.linear_proj.weight
662
+ module.decoder.layers.12.self_attention.linear_qkv.layer_norm_weight
663
+ module.decoder.layers.12.self_attention.linear_proj.weight
664
+ module.decoder.layers.11.mlp.router.weight
665
+ module.decoder.layers.11.pre_mlp_layernorm.weight
666
+ module.decoder.layers.11.self_attention.linear_proj.weight
667
+ module.decoder.layers.10.mlp.router.weight
668
+ module.decoder.layers.10.pre_mlp_layernorm.weight
669
+ module.decoder.layers.13.self_attention.linear_qkv.layer_norm_weight
670
+ module.decoder.layers.12.self_attention.linear_qkv.weight
671
+ module.decoder.layers.11.self_attention.linear_qkv.weight
672
+ module.decoder.layers.11.self_attention.linear_qkv.layer_norm_weight
673
+ module.decoder.layers.10.self_attention.linear_qkv.weight
674
+ Params for bucket 4 (50737152 elements, 50737152 padded size):
675
+ module.decoder.layers.9.pre_mlp_layernorm.weight
676
+ module.decoder.layers.9.self_attention.linear_qkv.weight
677
+ module.decoder.layers.8.mlp.router.weight
678
+ module.decoder.layers.8.pre_mlp_layernorm.weight
679
+ module.decoder.layers.7.mlp.router.weight
680
+ module.decoder.layers.7.pre_mlp_layernorm.weight
681
+ module.decoder.layers.10.self_attention.linear_qkv.layer_norm_weight
682
+ module.decoder.layers.9.mlp.router.weight
683
+ module.decoder.layers.10.self_attention.linear_proj.weight
684
+ module.decoder.layers.9.self_attention.linear_qkv.layer_norm_weight
685
+ module.decoder.layers.9.self_attention.linear_proj.weight
686
+ module.decoder.layers.8.self_attention.linear_qkv.weight
687
+ module.decoder.layers.8.self_attention.linear_qkv.layer_norm_weight
688
+ module.decoder.layers.8.self_attention.linear_proj.weight
689
+ module.decoder.layers.7.self_attention.linear_qkv.weight
690
+ Params for bucket 5 (50737152 elements, 50737152 padded size):
691
+ module.decoder.layers.7.self_attention.linear_qkv.layer_norm_weight
692
+ module.decoder.layers.6.self_attention.linear_qkv.weight
693
+ module.decoder.layers.6.pre_mlp_layernorm.weight
694
+ module.decoder.layers.6.mlp.router.weight
695
+ module.decoder.layers.6.self_attention.linear_proj.weight
696
+ module.decoder.layers.5.mlp.router.weight
697
+ module.decoder.layers.5.pre_mlp_layernorm.weight
698
+ module.decoder.layers.5.self_attention.linear_qkv.weight
699
+ module.decoder.layers.5.self_attention.linear_qkv.layer_norm_weight
700
+ module.decoder.layers.4.mlp.router.weight
701
+ module.decoder.layers.7.self_attention.linear_proj.weight
702
+ module.decoder.layers.6.self_attention.linear_qkv.layer_norm_weight
703
+ module.decoder.layers.5.self_attention.linear_proj.weight
704
+ module.decoder.layers.4.self_attention.linear_qkv.weight
705
+ module.decoder.layers.4.pre_mlp_layernorm.weight
706
+ Params for bucket 6 (50737152 elements, 50737152 padded size):
707
+ module.decoder.layers.3.pre_mlp_layernorm.weight
708
+ module.decoder.layers.4.self_attention.linear_proj.weight
709
+ module.decoder.layers.3.mlp.router.weight
710
+ module.decoder.layers.2.mlp.router.weight
711
+ module.decoder.layers.2.pre_mlp_layernorm.weight
712
+ module.decoder.layers.2.self_attention.linear_qkv.layer_norm_weight
713
+ module.decoder.layers.1.mlp.router.weight
714
+ module.decoder.layers.1.pre_mlp_layernorm.weight
715
+ module.decoder.layers.4.self_attention.linear_qkv.layer_norm_weight
716
+ module.decoder.layers.3.self_attention.linear_qkv.weight
717
+ module.decoder.layers.3.self_attention.linear_qkv.layer_norm_weight
718
+ module.decoder.layers.3.self_attention.linear_proj.weight
719
+ module.decoder.layers.2.self_attention.linear_qkv.weight
720
+ module.decoder.layers.2.self_attention.linear_proj.weight
721
+ module.decoder.layers.1.self_attention.linear_qkv.weight
722
+ Params for bucket 7 (124131328 elements, 124131328 padded size):
723
+ module.decoder.layers.0.pre_mlp_layernorm.weight
724
+ module.decoder.layers.1.self_attention.linear_proj.weight
725
+ module.decoder.layers.0.self_attention.linear_proj.weight
726
+ module.embedding.word_embeddings.weight
727
+ module.decoder.layers.1.self_attention.linear_qkv.layer_norm_weight
728
+ module.decoder.layers.0.self_attention.linear_qkv.weight
729
+ module.decoder.layers.0.mlp.router.weight
730
+ module.decoder.layers.0.self_attention.linear_qkv.layer_norm_weight
731
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 20
732
+ Params for bucket 1 (41943040 elements, 41943040 padded size):
733
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight7
734
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight5
735
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight3
736
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight1
737
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight6
738
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight4
739
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight2
740
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight6
741
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight4
742
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight2
743
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight0
744
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight7
745
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight5
746
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight3
747
+ Params for bucket 2 (41943040 elements, 41943040 padded size):
748
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight0
749
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight6
750
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight4
751
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight2
752
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight0
753
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight7
754
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight5
755
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight1
756
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight7
757
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight5
758
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight3
759
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight1
760
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight6
761
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight4
762
+ Params for bucket 3 (41943040 elements, 41943040 padded size):
763
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight3
764
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight1
765
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight7
766
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight5
767
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight3
768
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight1
769
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight6
770
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight0
771
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight2
772
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight6
773
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight4
774
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight2
775
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight0
776
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight7
777
+ Params for bucket 4 (41943040 elements, 41943040 padded size):
778
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight2
779
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight4
780
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight0
781
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight6
782
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight4
783
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight2
784
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight0
785
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight5
786
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight3
787
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight1
788
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight7
789
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight5
790
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight3
791
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight1
792
+ Params for bucket 5 (41943040 elements, 41943040 padded size):
793
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight7
794
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight3
795
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight5
796
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight1
797
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight6
798
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight4
799
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight6
800
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight4
801
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight2
802
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight0
803
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight7
804
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight5
805
+ Params for bucket 6 (41943040 elements, 41943040 padded size):
806
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight2
807
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight0
808
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight7
809
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight5
810
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight3
811
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight1
812
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight3
813
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight1
814
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight6
815
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight4
816
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight2
817
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight0
818
+ Params for bucket 7 (41943040 elements, 41943040 padded size):
819
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight7
820
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight5
821
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight3
822
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight1
823
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight7
824
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight5
825
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight3
826
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight4
827
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight6
828
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight2
829
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight0
830
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight6
831
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight4
832
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight2
833
+ Params for bucket 8 (41943040 elements, 41943040 padded size):
834
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight1
835
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight7
836
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight5
837
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight3
838
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight1
839
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight6
840
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight4
841
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight6
842
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight0
843
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight4
844
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight2
845
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight0
846
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight7
847
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight5
848
+ Params for bucket 9 (41943040 elements, 41943040 padded size):
849
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight7
850
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight0
851
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight2
852
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight5
853
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight3
854
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight1
855
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight6
856
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight3
857
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight1
858
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight6
859
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight4
860
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight2
861
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight0
862
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight7
863
+ Params for bucket 10 (41943040 elements, 41943040 padded size):
864
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight2
865
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight4
866
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight0
867
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight7
868
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight5
869
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight3
870
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight1
871
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight5
872
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight3
873
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight1
874
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight6
875
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight4
876
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight2
877
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight0
878
+ Params for bucket 11 (41943040 elements, 41943040 padded size):
879
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight6
880
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight4
881
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight2
882
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight0
883
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight6
884
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight4
885
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight7
886
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight3
887
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight5
888
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight1
889
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight7
890
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight5
891
+ Params for bucket 12 (41943040 elements, 41943040 padded size):
892
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight7
893
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight2
894
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight0
895
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight5
896
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight3
897
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight1
898
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight3
899
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight1
900
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight6
901
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight4
902
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight2
903
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight0
904
+ Params for bucket 13 (41943040 elements, 41943040 padded size):
905
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight4
906
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight6
907
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight2
908
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight0
909
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight7
910
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight5
911
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight3
912
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight7
913
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight5
914
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight3
915
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight1
916
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight6
917
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight4
918
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight2
919
+ Params for bucket 14 (41943040 elements, 41943040 padded size):
920
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight1
921
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight6
922
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight4
923
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight2
924
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight0
925
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight6
926
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight4
927
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight7
928
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight5
929
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight0
930
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight3
931
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight1
932
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight7
933
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight5
934
+ Params for bucket 15 (41943040 elements, 41943040 padded size):
935
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight0
936
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight2
937
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight6
938
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight4
939
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight2
940
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight0
941
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight7
942
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight3
943
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight7
944
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight1
945
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight5
946
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight3
947
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight1
948
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight6
949
+ Params for bucket 16 (41943040 elements, 41943040 padded size):
950
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight5
951
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight1
952
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight3
953
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight6
954
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight4
955
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight2
956
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight0
957
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight4
958
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight2
959
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight0
960
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight7
961
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight5
962
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight3
963
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight1
964
+ Params for bucket 17 (41943040 elements, 41943040 padded size):
965
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight4
966
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight6
967
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight2
968
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight0
969
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight6
970
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight4
971
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight7
972
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight3
973
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight5
974
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight1
975
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight7
976
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight5
977
+ Params for bucket 18 (41943040 elements, 41943040 padded size):
978
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight2
979
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight7
980
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight0
981
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight5
982
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight3
983
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight1
984
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight3
985
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight1
986
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight6
987
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight4
988
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight2
989
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight0
990
+ Params for bucket 19 (41943040 elements, 41943040 padded size):
991
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight7
992
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight6
993
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight4
994
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight2
995
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight7
996
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight5
997
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight2
998
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight3
999
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight5
1000
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight1
1001
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight0
1002
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight6
1003
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight4
1004
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight3
1005
+ Params for bucket 20 (8388608 elements, 8388608 padded size):
1006
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight1
1007
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight0
1008
+ 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 0x75daac5821b0>, config_logger_dir='')
1009
+ INFO:megatron.core.optimizer_param_scheduler:> learning rate decay style: cosine
1010
+ [after model, optimizer, and learning rate scheduler are built] datetime: 2026-09-21 08:36:00
1011
+ > building train, validation, and test datasets ...
1012
+ > datasets target sizes (minimum size):
1013
+ train: 96
1014
+ validation: 0
1015
+ test: 0
1016
+ INFO:megatron.core.datasets.blended_megatron_dataset_config:Let split_matrix = [(0, 1.0), None, None]
1017
+ > building train, validation, and test datasets for GPT ...
1018
+ INFO:megatron.core.datasets.blended_megatron_dataset_builder:Building GPTDataset splits with sizes=(96, 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 0x75daac582210>, 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)
1019
+ INFO:megatron.core.datasets.indexed_dataset:Load the _IndexReader from /workspace/data/wikitext/wikitext103_gpt2_text_document.idx
1020
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence lengths
1021
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence pointers
1022
+ INFO:megatron.core.datasets.indexed_dataset: Extract the document indices
1023
+ INFO:megatron.core.datasets.indexed_dataset:> total number of sequences: 1165029
1024
+ INFO:megatron.core.datasets.indexed_dataset:> total number of documents: 1801350
1025
+ INFO:megatron.core.datasets.gpt_dataset:Load the GPTDataset train indices
1026
+ INFO:megatron.core.datasets.gpt_dataset: Load the document index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-document_index.npy
1027
+ INFO:megatron.core.datasets.gpt_dataset: Load the sample index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-sample_index.npy
1028
+ INFO:megatron.core.datasets.gpt_dataset: Load the shuffle index from 466b63c71b42dae59ebdc25db6336051-GPTDataset-train-shuffle_index.npy
1029
+ INFO:megatron.core.datasets.gpt_dataset:> total number of samples: 29073
1030
+ > finished creating GPT datasets ...
1031
+ [after dataloaders are built] datetime: 2026-09-21 08:36:00
1032
+ done with setup ...
1033
+ (min, max) time across ranks (ms):
1034
+ model-and-optimizer-setup ......................: (300.76, 305.81)
1035
+ train/valid/test-data-iterators-setup ..........: (64.91, 68.07)
1036
+ training ...
1037
+ Overwriting rerun_state_machine.current_iteration from -1 to 0...
1038
+ [before the start of training step] datetime: 2026-09-21 08:36:00
1039
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1042
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1043
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1044
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1048
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1052
+ warnings.warn("The num_dispatched_tokens 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:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1056
+ warnings.warn("The num_dispatched_tokens 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:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1062
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1063
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1066
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1067
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1070
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1071
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1072
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1073
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1074
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1075
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1076
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1077
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1078
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1079
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1080
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1081
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1082
+ warnings.warn("The num_dispatched_tokens 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:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1086
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1087
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: 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
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1092
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1093
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1094
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1095
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1096
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1097
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1098
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1099
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1100
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1101
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1102
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1103
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1104
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1105
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1106
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1107
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1108
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1109
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1110
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1111
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1112
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1113
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1114
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1115
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1116
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1117
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1118
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1119
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1120
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1121
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1122
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1123
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1124
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1125
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1126
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1127
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1128
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1129
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1130
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1131
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1132
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1133
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1134
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1135
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1136
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1137
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1138
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1139
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1140
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1141
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1142
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1143
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1144
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1145
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1146
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1147
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1148
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1149
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
1150
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
1151
+ [2026-09-21 08:36:38] iteration 1/ 12 | consumed samples: 8 | elapsed time per iteration (ms): 38190.1 | throughput per GPU (TFLOP/s/GPU): 0.8 | learning rate: 2.000000E-05 | global batch size: 8 | lm loss: 1.118499E+01 | loss scale: 1.0 | grad norm: 35.477 | number of skipped iterations: 0 | number of nan iterations: 0 |Number of parameters in transformer block in billions: 6.71
1152
+
1153
+ Number of parameters in embedding layers in billions: 0.21
1154
+ Total number of parameters in billions: 6.92
1155
+ Number of parameters in most loaded shard in billions: 6.9171
1156
+ Theoretical memory footprints: weight and optimizer=49474.72 MB
1157
+ [Rank 0] (after 1 iterations) memory (MB) | allocated: 19780.87158203125 | max allocated: 23491.14599609375 | reserved: 24330.0 | max reserved: 24330.0
1158
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1159
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1160
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1161
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1162
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1163
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1164
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1165
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1166
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1167
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1168
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1169
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1170
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1171
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1172
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
1173
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
1174
+ [2026-09-21 08:36:38] iteration 2/ 12 | consumed samples: 16 | elapsed time per iteration (ms): 279.1 | throughput per GPU (TFLOP/s/GPU): 115.4 | learning rate: 4.000000E-05 | global batch size: 8 | lm loss: 1.120955E+01 | loss scale: 1.0 | grad norm: 36.283 | number of skipped iterations: 0 | number of nan iterations: 0 |
1175
+ [2026-09-21 08:36:39] iteration 3/ 12 | consumed samples: 24 | elapsed time per iteration (ms): 539.0 | throughput per GPU (TFLOP/s/GPU): 59.8 | learning rate: 6.000000E-05 | global batch size: 8 | lm loss: 1.006648E+01 | loss scale: 1.0 | grad norm: 101.506 | number of skipped iterations: 0 | number of nan iterations: 0 |
1176
+ Capture CUDA graph for training!!!
1177
+ CUDA graph capture done!!!
1178
+ [2026-09-21 08:36:40] iteration 4/ 12 | consumed samples: 32 | elapsed time per iteration (ms): 1066.3 | throughput per GPU (TFLOP/s/GPU): 30.2 | learning rate: 8.000000E-05 | global batch size: 8 | lm loss: 1.023731E+01 | loss scale: 1.0 | grad norm: 29.927 | number of skipped iterations: 0 | number of nan iterations: 0 |
1179
+ [2026-09-21 08:36:40] iteration 5/ 12 | consumed samples: 40 | elapsed time per iteration (ms): 110.5 | throughput per GPU (TFLOP/s/GPU): 291.5 | learning rate: 1.000000E-04 | global batch size: 8 | lm loss: 1.077410E+01 | loss scale: 1.0 | grad norm: 21.335 | number of skipped iterations: 0 | number of nan iterations: 0 |
1180
+ [2026-09-21 08:36:40] iteration 6/ 12 | consumed samples: 48 | elapsed time per iteration (ms): 110.5 | throughput per GPU (TFLOP/s/GPU): 291.5 | learning rate: 7.750000E-05 | global batch size: 8 | lm loss: 9.669871E+00 | loss scale: 1.0 | grad norm: 8.499 | number of skipped iterations: 0 | number of nan iterations: 0 |
1181
+ [2026-09-21 08:36:40] iteration 7/ 12 | consumed samples: 56 | elapsed time per iteration (ms): 113.1 | throughput per GPU (TFLOP/s/GPU): 284.8 | learning rate: 3.250000E-05 | global batch size: 8 | lm loss: 9.129749E+00 | loss scale: 1.0 | grad norm: 15.329 | number of skipped iterations: 0 | number of nan iterations: 0 |
1182
+ [2026-09-21 08:36:40] iteration 8/ 12 | consumed samples: 64 | elapsed time per iteration (ms): 110.9 | throughput per GPU (TFLOP/s/GPU): 290.4 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.839231E+00 | loss scale: 1.0 | grad norm: 6.966 | number of skipped iterations: 0 | number of nan iterations: 0 |
1183
+ [2026-09-21 08:36:41] iteration 9/ 12 | consumed samples: 72 | elapsed time per iteration (ms): 109.8 | throughput per GPU (TFLOP/s/GPU): 293.6 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.722826E+00 | loss scale: 1.0 | grad norm: 4.102 | number of skipped iterations: 0 | number of nan iterations: 0 |
1184
+ [2026-09-21 08:36:41] iteration 10/ 12 | consumed samples: 80 | elapsed time per iteration (ms): 110.3 | throughput per GPU (TFLOP/s/GPU): 292.1 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.658896E+00 | loss scale: 1.0 | grad norm: 3.601 | number of skipped iterations: 0 | number of nan iterations: 0 |
1185
+ [2026-09-21 08:36:41] iteration 11/ 12 | consumed samples: 88 | elapsed time per iteration (ms): 109.7 | throughput per GPU (TFLOP/s/GPU): 293.7 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.643707E+00 | loss scale: 1.0 | grad norm: 3.521 | number of skipped iterations: 0 | number of nan iterations: 0 |
1186
+ [2026-09-21 08:36:41] iteration 12/ 12 | consumed samples: 96 | elapsed time per iteration (ms): 108.8 | throughput per GPU (TFLOP/s/GPU): 296.0 | learning rate: 1.000000E-05 | global batch size: 8 | lm loss: 8.591064E+00 | loss scale: 1.0 | grad norm: 3.404 | number of skipped iterations: 0 | number of nan iterations: 0 |
1187
+ [after training is done] datetime: 2026-09-21 08:36:41
1188
+ [rank0]:[W921 08:36:42.517299038 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())
1189
+ W0921 08:36:45.932000 686861 torch/distributed/elastic/agent/server/api.py:719] Received 15 death signal, shutting down workers
1190
+ W0921 08:36:45.932000 686861 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 686951 closing signal SIGTERM
1191
+ W0921 08:36:45.933000 686861 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 686953 closing signal SIGTERM
1192
+ W0921 08:36:45.933000 686861 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 686955 closing signal SIGTERM
1193
+ Traceback (most recent call last):
1194
+ File "/usr/local/bin/torchrun", line 8, in <module>
1195
+ sys.exit(main())
1196
+ ^^^^^^
1197
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper
1198
+ return f(*args, **kwargs)
1199
+ ^^^^^^^^^^^^^^^^^^
1200
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 892, in main
1201
+ run(args)
1202
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 883, in run
1203
+ elastic_launch(
1204
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 139, in __call__
1205
+ return launch_agent(self._config, self._entrypoint, list(args))
1206
+ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1207
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 261, in launch_agent
1208
+ result = agent.run()
1209
+ ^^^^^^^^^^^
1210
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper
1211
+ result = f(*args, **kwargs)
1212
+ ^^^^^^^^^^^^^^^^^^
1213
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run
1214
+ result = self._invoke_run(role)
1215
+ ^^^^^^^^^^^^^^^^^^^^^^
1216
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run
1217
+ time.sleep(monitor_interval)
1218
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler
1219
+ raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
1220
+ torch.distributed.elastic.multiprocessing.api.SignalException: Process 686861 got signal: 15
granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/results.tsv ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ config backend policy slots seq gbs cap recompute rep status iter_ms_mean iter_ms_min peak_mib run_dir
2
+ OLMOE_1B_7B_L16 echo matched 8 4096 8 1.08984375 none 1 OK 110.13 108.40 43043 /workspace/granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep1
3
+ OLMOE_1B_7B_L16 echo matched 8 4096 8 1.08984375 none 2 OK 110.79 108.40 43043 /workspace/granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep2
4
+ OLMOE_1B_7B_L16 echo matched 8 4096 8 1.08984375 none 3 OK 110.68 108.40 43043 /workspace/granularity_next_b200/tierb_models/attempts/OLMOE_1B_7B_L16/echo/tierb/tierb_gbs8/OLMOE_1B_7B_L16__echo__matched__seq4096__gbs8__cap1p08984375/rep3