AbhishekGhosh1998 commited on
Commit
05d3ca0
·
verified ·
1 Parent(s): 1c3d168

granularity results tiera-rerun-fixed-moonep/echo/ponly @ 2026-08-30T06:44:12Z from sp-aiinfra-interactive-aghosh-7jtjl-8xb200-moe-cg-0

Browse files
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep1/mem_samples.csv CHANGED
@@ -22,3 +22,219 @@
22
  5, 1
23
  6, 1
24
  7, 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, 629
44
+ 3, 629
45
+ 4, 629
46
+ 5, 629
47
+ 6, 629
48
+ 7, 629
49
+ 0, 925
50
+ 1, 913
51
+ 2, 909
52
+ 3, 919
53
+ 4, 915
54
+ 5, 941
55
+ 6, 927
56
+ 7, 925
57
+ 0, 2007
58
+ 1, 2007
59
+ 2, 2007
60
+ 3, 2007
61
+ 4, 2007
62
+ 5, 2007
63
+ 6, 2007
64
+ 7, 2007
65
+ 0, 12399
66
+ 1, 12397
67
+ 2, 12397
68
+ 3, 12397
69
+ 4, 12397
70
+ 5, 12397
71
+ 6, 12397
72
+ 7, 12397
73
+ 0, 12927
74
+ 1, 12923
75
+ 2, 12923
76
+ 3, 12923
77
+ 4, 12925
78
+ 5, 12927
79
+ 6, 12923
80
+ 7, 12923
81
+ 0, 13199
82
+ 1, 13197
83
+ 2, 13197
84
+ 3, 13197
85
+ 4, 13197
86
+ 5, 13197
87
+ 6, 13197
88
+ 7, 13197
89
+ 0, 13813
90
+ 1, 13811
91
+ 2, 13811
92
+ 3, 13811
93
+ 4, 13811
94
+ 5, 13811
95
+ 6, 13811
96
+ 7, 13811
97
+ 0, 13837
98
+ 1, 13835
99
+ 2, 13835
100
+ 3, 13835
101
+ 4, 13835
102
+ 5, 13835
103
+ 6, 13835
104
+ 7, 13835
105
+ 0, 13837
106
+ 1, 13835
107
+ 2, 13835
108
+ 3, 13835
109
+ 4, 13835
110
+ 5, 13835
111
+ 6, 13835
112
+ 7, 13835
113
+ 0, 14483
114
+ 1, 14481
115
+ 2, 14481
116
+ 3, 14481
117
+ 4, 14481
118
+ 5, 14481
119
+ 6, 14481
120
+ 7, 14481
121
+ 0, 15009
122
+ 1, 15007
123
+ 2, 15007
124
+ 3, 15007
125
+ 4, 15007
126
+ 5, 15007
127
+ 6, 15007
128
+ 7, 15007
129
+ 0, 15317
130
+ 1, 15315
131
+ 2, 15315
132
+ 3, 15315
133
+ 4, 15315
134
+ 5, 15315
135
+ 6, 15315
136
+ 7, 15315
137
+ 0, 16295
138
+ 1, 16293
139
+ 2, 16293
140
+ 3, 16293
141
+ 4, 16293
142
+ 5, 16293
143
+ 6, 16293
144
+ 7, 16293
145
+ 0, 29613
146
+ 1, 29961
147
+ 2, 32899
148
+ 3, 29961
149
+ 4, 29921
150
+ 5, 30709
151
+ 6, 28895
152
+ 7, 28871
153
+ 0, 34621
154
+ 1, 34599
155
+ 2, 34599
156
+ 3, 34599
157
+ 4, 34561
158
+ 5, 34599
159
+ 6, 34599
160
+ 7, 34599
161
+ 0, 34621
162
+ 1, 34599
163
+ 2, 34599
164
+ 3, 34599
165
+ 4, 34561
166
+ 5, 34599
167
+ 6, 34599
168
+ 7, 34599
169
+ 0, 34621
170
+ 1, 34599
171
+ 2, 34599
172
+ 3, 34599
173
+ 4, 34561
174
+ 5, 34599
175
+ 6, 34599
176
+ 7, 34599
177
+ 0, 34621
178
+ 1, 34599
179
+ 2, 34599
180
+ 3, 34599
181
+ 4, 34561
182
+ 5, 34599
183
+ 6, 34599
184
+ 7, 34599
185
+ 0, 37065
186
+ 1, 37039
187
+ 2, 37057
188
+ 3, 37037
189
+ 4, 37033
190
+ 5, 37071
191
+ 6, 37049
192
+ 7, 37085
193
+ 0, 37801
194
+ 1, 37793
195
+ 2, 37799
196
+ 3, 37839
197
+ 4, 37749
198
+ 5, 37829
199
+ 6, 37825
200
+ 7, 37795
201
+ 0, 38559
202
+ 1, 38537
203
+ 2, 38537
204
+ 3, 38537
205
+ 4, 38499
206
+ 5, 38537
207
+ 6, 38537
208
+ 7, 38537
209
+ 0, 39173
210
+ 1, 39163
211
+ 2, 39169
212
+ 3, 39209
213
+ 4, 39151
214
+ 5, 39195
215
+ 6, 39185
216
+ 7, 39207
217
+ 0, 53279
218
+ 1, 53101
219
+ 2, 53103
220
+ 3, 53103
221
+ 4, 52755
222
+ 5, 53101
223
+ 6, 53101
224
+ 7, 53103
225
+ 0, 53279
226
+ 1, 53101
227
+ 2, 53103
228
+ 3, 53103
229
+ 4, 52755
230
+ 5, 53101
231
+ 6, 53101
232
+ 7, 53103
233
+ 0, 1045
234
+ 1, 53101
235
+ 2, 2328
236
+ 3, 728
237
+ 4, 728
238
+ 5, 728
239
+ 6, 728
240
+ 7, 728
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep1/train.log CHANGED
@@ -10,3 +10,987 @@ W0830 06:40:27.985000 640669 torch/distributed/run.py:766]
10
  W0830 06:40:27.985000 640669 torch/distributed/run.py:766] *****************************************
11
  W0830 06:40:27.985000 640669 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
  W0830 06:40:27.985000 640669 torch/distributed/run.py:766] *****************************************
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
10
  W0830 06:40:27.985000 640669 torch/distributed/run.py:766] *****************************************
11
  W0830 06:40:27.985000 640669 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
  W0830 06:40:27.985000 640669 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
+ 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
18
+ 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
19
+ WARNING: overriding default arguments for tokenizer_type:GPT2BPETokenizer with tokenizer_type:GPT2BPETokenizer
20
+ Number of virtual stages per pipeline stage: None
21
+ accumulate and all-reduce gradients in fp32 for bfloat16 data type.
22
+ using torch.bfloat16 for parameters ...
23
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
24
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
25
+ ------------------------ arguments ------------------------
26
+ account_for_embedding_in_pipeline_split ......... False
27
+ account_for_loss_in_pipeline_split .............. False
28
+ accumulate_allreduce_grads_in_fp32 .............. True
29
+ adam_beta1 ...................................... 0.9
30
+ adam_beta2 ...................................... 0.999
31
+ adam_eps ........................................ 1e-08
32
+ add_bias_linear ................................. False
33
+ add_position_embedding .......................... False
34
+ add_qkv_bias .................................... False
35
+ adlr_autoresume ................................. False
36
+ adlr_autoresume_interval ........................ 1000
37
+ align_grad_reduce ............................... True
38
+ align_param_gather .............................. False
39
+ app_tag_run_name ................................ None
40
+ app_tag_run_version ............................. 0.0.0
41
+ apply_layernorm_1p .............................. False
42
+ apply_query_key_layer_scaling ................... False
43
+ apply_residual_connection_post_layernorm ........ False
44
+ apply_rope_fusion ............................... True
45
+ async_save ...................................... None
46
+ async_tensor_model_parallel_allreduce ........... True
47
+ attention_backend ............................... AttnBackend.auto
48
+ attention_dropout ............................... 0.0
49
+ attention_softmax_in_fp32 ....................... False
50
+ auto_detect_ckpt_format ......................... False
51
+ barrier_with_L1_time ............................ True
52
+ bert_binary_head ................................ True
53
+ bert_embedder_type .............................. megatron
54
+ bert_load ....................................... None
55
+ bf16 ............................................ True
56
+ bias_dropout_fusion ............................. True
57
+ bias_gelu_fusion ................................ False
58
+ bias_swiglu_fusion .............................. True
59
+ biencoder_projection_dim ........................ 0
60
+ biencoder_shared_query_context_model ............ False
61
+ block_data_path ................................. None
62
+ calc_ft_timeouts ................................ False
63
+ calculate_per_token_loss ........................ False
64
+ check_for_large_grads ........................... False
65
+ check_for_nan_in_loss_and_grad .................. False
66
+ check_for_spiky_loss ............................ False
67
+ check_weight_hash_across_dp_replicas_interval ... None
68
+ ckpt_assume_constant_structure .................. False
69
+ ckpt_convert_format ............................. None
70
+ ckpt_convert_save ............................... None
71
+ ckpt_convert_update_legacy_dist_opt_format ...... False
72
+ ckpt_format ..................................... torch_dist
73
+ ckpt_fully_parallel_load ........................ False
74
+ ckpt_fully_parallel_save ........................ True
75
+ ckpt_fully_parallel_save_deprecated ............. False
76
+ ckpt_step ....................................... None
77
+ classes_fraction ................................ 1.0
78
+ clip_grad ....................................... 1.0
79
+ clone_scatter_output_in_embedding ............... True
80
+ config_logger_dir ...............................
81
+ consumed_train_samples .......................... 0
82
+ consumed_valid_samples .......................... 0
83
+ context_parallel_size ........................... 1
84
+ cp_comm_type .................................... ['p2p']
85
+ create_attention_mask_in_dataloader ............. True
86
+ cross_entropy_fusion_impl ....................... native
87
+ cross_entropy_loss_fusion ....................... False
88
+ cuda_graph_scope ................................ full_iteration
89
+ cuda_graph_warmup_steps ......................... 3
90
+ data_args_path .................................. None
91
+ data_cache_path ................................. None
92
+ data_parallel_random_init ....................... False
93
+ data_parallel_sharding_strategy ................. no_shard
94
+ data_parallel_size .............................. 8
95
+ data_path ....................................... ['/workspace/data/wikitext/wikitext103_gpt2_text_document']
96
+ data_per_class_fraction ......................... 1.0
97
+ data_sharding ................................... True
98
+ dataloader_type ................................. single
99
+ ddp_average_in_collective ....................... False
100
+ ddp_bucket_size ................................. None
101
+ ddp_num_buckets ................................. None
102
+ ddp_pad_buckets_for_high_nccl_busbw ............. False
103
+ decoder_first_pipeline_num_layers ............... None
104
+ decoder_last_pipeline_num_layers ................ None
105
+ decoder_num_layers .............................. None
106
+ decoder_seq_length .............................. None
107
+ decoupled_lr .................................... None
108
+ decoupled_min_lr ................................ None
109
+ decrease_batch_size_if_needed ................... False
110
+ defer_embedding_wgrad_compute ................... False
111
+ delay_wgrad_compute ............................. False
112
+ deprecated_use_mcore_models ..................... True
113
+ deterministic_mode .............................. False
114
+ dino_bottleneck_size ............................ 256
115
+ dino_freeze_last_layer .......................... 1
116
+ dino_head_hidden_size ........................... 2048
117
+ dino_local_crops_number ......................... 10
118
+ dino_local_img_size ............................. 96
119
+ dino_norm_last_layer ............................ False
120
+ dino_teacher_temp ............................... 0.07
121
+ dino_warmup_teacher_temp ........................ 0.04
122
+ dino_warmup_teacher_temp_epochs ................. 30
123
+ disable_bf16_reduced_precision_matmul ........... False
124
+ disable_mamba_mem_eff_path ...................... False
125
+ disable_straggler_on_startup .................... False
126
+ dist_ckpt_format_deprecated ..................... None
127
+ dist_ckpt_strictness ............................ assume_ok_unexpected
128
+ distribute_saved_activations .................... False
129
+ distributed_backend ............................. nccl
130
+ distributed_timeout_minutes ..................... 10
131
+ embedding_init_method_std ....................... None
132
+ embedding_path .................................. None
133
+ empty_unused_memory_level ....................... 0
134
+ enable_cuda_graph ............................... True
135
+ enable_experimental ............................. False
136
+ enable_ft_package ............................... False
137
+ enable_gloo_process_groups ...................... True
138
+ enable_msc ...................................... True
139
+ enable_one_logger ............................... True
140
+ encoder_num_layers .............................. 16
141
+ encoder_seq_length .............................. 8192
142
+ end_weight_decay ................................ 0.1
143
+ eod_mask_loss ................................... False
144
+ error_injection_rate ............................ 0
145
+ error_injection_type ............................ transient_error
146
+ eval_interval ................................... 1000000
147
+ eval_iters ...................................... 0
148
+ evidence_data_path .............................. None
149
+ exit_duration_in_mins ........................... None
150
+ exit_interval ................................... None
151
+ exit_on_missing_checkpoint ...................... False
152
+ exit_signal_handler ............................. False
153
+ exp_avg_dtype ................................... torch.float32
154
+ exp_avg_sq_dtype ................................ torch.float32
155
+ expert_model_parallel_size ...................... 8
156
+ expert_tensor_parallel_size ..................... 1
157
+ export_force_local_attention .................... False
158
+ export_kd_cfg ................................... None
159
+ export_kd_teacher_ckpt_format ................... None
160
+ export_kd_teacher_load .......................... None
161
+ export_kv_cache_quant ........................... False
162
+ export_legacy_megatron .......................... False
163
+ export_model_type ............................... GPTModel
164
+ export_moe_apply_probs_on_input ................. False
165
+ export_qk_l2_norm ............................... False
166
+ export_quant_cfg ................................ None
167
+ export_real_quant_cfg ........................... None
168
+ export_te_mcore_model ........................... False
169
+ external_cuda_graph ............................. False
170
+ ffn_hidden_size ................................. 14336
171
+ finetune ........................................ False
172
+ finetune_data_split ............................. train
173
+ finetune_hf_dataset ............................. None
174
+ first_last_layers_bf16 .......................... False
175
+ flash_decode .................................... False
176
+ fp16 ............................................ False
177
+ fp16_lm_cross_entropy ........................... False
178
+ fp32_residual_connection ........................ False
179
+ fp8 ............................................. e4m3
180
+ fp8_amax_compute_algo ........................... most_recent
181
+ fp8_amax_history_len ............................ 1
182
+ fp8_interval .................................... 1
183
+ fp8_margin ...................................... 0
184
+ fp8_param_gather ................................ False
185
+ fp8_recipe ...................................... mxfp8
186
+ fp8_wgrad ....................................... True
187
+ fsdp_double_buffer .............................. False
188
+ global_batch_size ............................... 16
189
+ grad_reduce_in_bf16 ............................. False
190
+ gradient_accumulation_fusion .................... True
191
+ gradient_reduce_div_fusion ...................... True
192
+ group_query_attention ........................... True
193
+ head_lr_mult .................................... 1.0
194
+ heterogeneous_layers_config_encoded_json ........ None
195
+ heterogeneous_layers_config_path ................ None
196
+ hidden_dropout .................................. 0.0
197
+ hidden_size ..................................... 4096
198
+ hierarchical_context_parallel_sizes ............. None
199
+ high_priority_stream_groups ..................... []
200
+ hybrid_attention_ratio .......................... 0.0
201
+ hybrid_mlp_ratio ................................ 0.0
202
+ hybrid_override_pattern ......................... None
203
+ hysteresis ...................................... 2
204
+ ict_head_size ................................... None
205
+ ict_load ........................................ None
206
+ img_h ........................................... 224
207
+ img_w ........................................... 224
208
+ indexer_batch_size .............................. 128
209
+ indexer_log_interval ............................ 1000
210
+ inference_batch_times_seqlen_threshold .......... -1
211
+ inference_dynamic_batching ...................... False
212
+ inference_dynamic_batching_buffer_guaranteed_fraction 0.2
213
+ inference_dynamic_batching_buffer_overflow_factor None
214
+ inference_dynamic_batching_buffer_size_gb ....... 40.0
215
+ inference_dynamic_batching_chunk_size ........... 256
216
+ inference_dynamic_batching_max_requests_override None
217
+ inference_dynamic_batching_max_tokens_override .. None
218
+ inference_dynamic_batching_num_cuda_graphs ...... 16
219
+ inference_max_batch_size ........................ 8
220
+ inference_max_seq_length ........................ 2560
221
+ inference_rng_tracker ........................... False
222
+ init_method_std ................................. 0.01
223
+ init_method_xavier_uniform ...................... False
224
+ init_model_with_meta_device ..................... False
225
+ initial_loss_scale .............................. 4294967296
226
+ inprocess_active_world_size ..................... 8
227
+ inprocess_barrier_timeout ....................... 120
228
+ inprocess_completion_timeout .................... 120
229
+ inprocess_empty_cuda_cache ...................... False
230
+ inprocess_granularity ........................... node
231
+ inprocess_hard_timeout .......................... 90
232
+ inprocess_heartbeat_interval .................... 30
233
+ inprocess_heartbeat_timeout ..................... 60
234
+ inprocess_last_call_wait ........................ 1
235
+ inprocess_max_iterations ........................ None
236
+ inprocess_monitor_process_interval .............. 1.0
237
+ inprocess_monitor_thread_interval ............... 1.0
238
+ inprocess_progress_watchdog_interval ............ 1.0
239
+ inprocess_restart ............................... False
240
+ inprocess_soft_timeout .......................... 60
241
+ inprocess_termination_grace_time ................ 1
242
+ is_hybrid_model ................................. False
243
+ iter_per_epoch .................................. 1250
244
+ iterations_to_skip .............................. []
245
+ keep_fp8_transpose_cache_when_using_custom_fsdp . False
246
+ kitchen_config_file ............................. None
247
+ kitchen_recipe_number ........................... None
248
+ kv_channels ..................................... 128
249
+ kv_lora_rank .................................... 32
250
+ lazy_mpu_init ................................... None
251
+ load ............................................ None
252
+ load_main_params_from_ckpt ...................... None
253
+ load_model_opt_format ........................... False
254
+ local_rank ...................................... 0
255
+ log_energy ...................................... False
256
+ log_interval .................................... 1
257
+ log_loss_scale_to_tensorboard ................... True
258
+ log_memory_to_tensorboard ....................... False
259
+ log_num_zeros_in_grad ........................... False
260
+ log_params_norm ................................. False
261
+ log_progress .................................... False
262
+ log_straggler ................................... False
263
+ log_throughput .................................. True
264
+ log_timers_to_tensorboard ....................... False
265
+ log_validation_ppl_to_tensorboard ............... False
266
+ log_world_size_to_tensorboard ................... False
267
+ logging_level ................................... None
268
+ loss_scale ...................................... None
269
+ loss_scale_window ............................... 1000
270
+ lr .............................................. 0.0001
271
+ lr_decay_iters .................................. 8
272
+ lr_decay_samples ................................ None
273
+ lr_decay_style .................................. cosine
274
+ lr_warmup_fraction .............................. None
275
+ lr_warmup_init .................................. 0.0
276
+ lr_warmup_iters ................................. 5
277
+ lr_warmup_samples ............................... 0
278
+ lr_wsd_decay_iters .............................. None
279
+ lr_wsd_decay_samples ............................ None
280
+ lr_wsd_decay_style .............................. exponential
281
+ main_grads_dtype ................................ torch.float32
282
+ main_params_dtype ............................... torch.float32
283
+ make_vocab_size_divisible_by .................... 128
284
+ mamba_head_dim .................................. 64
285
+ mamba_num_groups ................................ 8
286
+ mamba_num_heads ................................. None
287
+ mamba_state_dim ................................. 128
288
+ manual_gc ....................................... False
289
+ manual_gc_eval .................................. True
290
+ manual_gc_interval .............................. 0
291
+ mask_factor ..................................... 1.0
292
+ mask_prob ....................................... 0.15
293
+ mask_type ....................................... random
294
+ masked_softmax_fusion ........................... False
295
+ max_position_embeddings ......................... 32768
296
+ max_tokens_to_oom ............................... 12000
297
+ memory_snapshot_path ............................ snapshot.pickle
298
+ merge_file ...................................... /workspace/data/wikitext/gpt2-merges.txt
299
+ micro_batch_size ................................ 1
300
+ microbatch_group_size_per_vp_stage .............. None
301
+ mid_level_dataset_surplus ....................... 0.005
302
+ min_loss_scale .................................. 1.0
303
+ min_lr .......................................... 1e-05
304
+ mlp_chunks_for_prefill .......................... 1
305
+ mmap_bin_files .................................. True
306
+ mock_data ....................................... False
307
+ moe_apply_probs_on_input ........................ False
308
+ moe_aux_loss_coeff .............................. 0.01
309
+ moe_deepep_num_sms .............................. 20
310
+ moe_echo_enable_random_offloading ............... False
311
+ moe_echo_expert_dispatch_overlap ................ True
312
+ moe_echo_expert_dispatcher_type ................. hybridep
313
+ moe_echo_recompute_expert_dispatch .............. False
314
+ moe_enable_deepep ............................... False
315
+ moe_enable_echo ................................. True
316
+ moe_enable_hybridep ............................. True
317
+ moe_expert_capacity_factor ...................... None
318
+ moe_extended_tp ................................. False
319
+ moe_ffn_hidden_size ............................. 1536
320
+ moe_grouped_gemm ................................ True
321
+ moe_hybridep_num_sms ............................ 16
322
+ moe_input_jitter_eps ............................ None
323
+ moe_layer_freq .................................. 1
324
+ moe_layer_recompute ............................. False
325
+ moe_num_echo_experts ............................ 8
326
+ moe_pad_expert_input_to_capacity ................ False
327
+ moe_per_layer_logging ........................... False
328
+ moe_permute_fusion .............................. False
329
+ moe_received_token_capacity ..................... 2.40625
330
+ moe_router_bias_update_rate ..................... 0.001
331
+ moe_router_dtype ................................ fp32
332
+ moe_router_enable_expert_bias ................... False
333
+ moe_router_force_load_balancing ................. False
334
+ moe_router_group_topk ........................... None
335
+ moe_router_load_balancing_type .................. aux_loss
336
+ moe_router_num_groups ........................... None
337
+ moe_router_padding_for_fp8 ...................... False
338
+ moe_router_pre_softmax .......................... False
339
+ moe_router_score_function ....................... softmax
340
+ moe_router_topk ................................. 2
341
+ moe_router_topk_scaling_factor .................. None
342
+ moe_shared_expert_intermediate_size ............. None
343
+ moe_shared_expert_overlap ....................... False
344
+ moe_token_dispatcher_type ....................... flex
345
+ moe_token_drop_policy ........................... probs
346
+ moe_upcycling_granularity ....................... 1
347
+ moe_use_device_initiated_grouped_gemm ........... True
348
+ moe_use_legacy_grouped_gemm ..................... False
349
+ moe_use_upcycling ............................... False
350
+ moe_z_loss_coeff ................................ None
351
+ mrope_section ................................... None
352
+ mscale .......................................... 1.0
353
+ mscale_all_dim .................................. 0.0
354
+ mtp_loss_scaling_factor ......................... 0.1
355
+ mtp_num_layers .................................. None
356
+ multi_latent_attention .......................... False
357
+ nccl_all_reduce_for_prefill ..................... False
358
+ nccl_communicator_config_path ................... None
359
+ nccl_ub ......................................... False
360
+ no_load_optim ................................... None
361
+ no_load_rng ..................................... None
362
+ no_persist_layer_norm ........................... False
363
+ no_rope_freq .................................... None
364
+ no_save_optim ................................... None
365
+ no_save_rng ..................................... None
366
+ non_persistent_ckpt_type ........................ None
367
+ non_persistent_global_ckpt_dir .................. None
368
+ non_persistent_local_ckpt_algo .................. fully_parallel
369
+ non_persistent_local_ckpt_dir ................... None
370
+ non_persistent_save_interval .................... None
371
+ norm_epsilon .................................... 1e-05
372
+ normalization ................................... RMSNorm
373
+ num_attention_heads ............................. 32
374
+ num_channels .................................... 3
375
+ num_classes ..................................... 1000
376
+ num_dataset_builder_threads ..................... 1
377
+ num_distributed_optimizer_instances ............. 1
378
+ num_experts ..................................... 8
379
+ num_layers ...................................... 16
380
+ num_layers_at_end_in_bf16 ....................... 1
381
+ num_layers_at_start_in_bf16 ..................... 1
382
+ num_layers_per_virtual_pipeline_stage ........... None
383
+ num_query_groups ................................ 8
384
+ num_virtual_stages_per_pipeline_rank ............ None
385
+ num_workers ..................................... 2
386
+ object_storage_cache_path ....................... None
387
+ one_logger_async ................................ False
388
+ one_logger_project .............................. megatron-lm
389
+ one_logger_run_name ............................. None
390
+ onnx_safe ....................................... None
391
+ openai_gelu ..................................... False
392
+ optimizer ....................................... adam
393
+ optimizer_cpu_offload ........................... False
394
+ optimizer_offload_fraction ...................... 1.0
395
+ output_bert_embeddings .......................... False
396
+ overlap_cpu_optimizer_d2h_h2d ................... False
397
+ overlap_grad_reduce ............................. True
398
+ overlap_moe_expert_parallel_comm ................ False
399
+ overlap_p2p_comm ................................ False
400
+ overlap_p2p_comm_warmup_flush ................... False
401
+ overlap_param_gather ............................ True
402
+ overlap_param_gather_with_optimizer_step ........ False
403
+ override_opt_param_scheduler .................... False
404
+ params_dtype .................................... torch.bfloat16
405
+ patch_dim ....................................... 16
406
+ per_split_data_args_path ........................ None
407
+ perform_initialization .......................... True
408
+ pin_cpu_grads ................................... True
409
+ pin_cpu_params .................................. True
410
+ pipeline_model_parallel_comm_backend ............ None
411
+ pipeline_model_parallel_layout .................. None
412
+ pipeline_model_parallel_size .................... 1
413
+ position_embedding_type ......................... rope
414
+ pretrained_checkpoint ........................... None
415
+ profile ......................................... False
416
+ profile_ranks ................................... [0]
417
+ profile_step_end ................................ 12
418
+ profile_step_start .............................. 10
419
+ q_lora_rank ..................................... None
420
+ qk_head_dim ..................................... 128
421
+ qk_l2_norm ...................................... False
422
+ qk_layernorm .................................... False
423
+ qk_pos_emb_head_dim ............................. 64
424
+ query_in_block_prob ............................. 0.1
425
+ rampup_batch_size ............................... None
426
+ rank ............................................ 0
427
+ recompute_granularity ........................... None
428
+ recompute_method ................................ None
429
+ recompute_modules ............................... None
430
+ recompute_num_layers ............................ None
431
+ record_memory_history ........................... False
432
+ relative_attention_max_distance ................. 128
433
+ relative_attention_num_buckets .................. 32
434
+ replication ..................................... False
435
+ replication_factor .............................. 2
436
+ replication_jump ................................ None
437
+ rerun_mode ...................................... validate_results
438
+ reset_attention_mask ............................ False
439
+ reset_position_ids .............................. False
440
+ result_rejected_tracker_filename ................ None
441
+ retriever_report_topk_accuracies ................ []
442
+ retriever_score_scaling ......................... False
443
+ retriever_seq_length ............................ 256
444
+ retro_add_retriever ............................. False
445
+ retro_attention_gate ............................ 1
446
+ retro_cyclic_train_iters ........................ None
447
+ retro_encoder_attention_dropout ................. 0.1
448
+ retro_encoder_hidden_dropout .................... 0.1
449
+ retro_encoder_layers ............................ 2
450
+ retro_num_neighbors ............................. 2
451
+ retro_num_retrieved_chunks ...................... 2
452
+ retro_project_dir ............................... None
453
+ retro_verify_neighbor_count ..................... True
454
+ reuse_grad_buf_for_mxfp8_param_ag ............... False
455
+ rope_scaling_factor ............................. 8.0
456
+ rope_type ....................................... None
457
+ rotary_base ..................................... 1000000
458
+ rotary_interleaved .............................. False
459
+ rotary_percent .................................. 1.0
460
+ rotary_scaling_factor ........................... 1.0
461
+ rotary_seq_len_interpolation_factor ............. None
462
+ run_workload_inspector_server ................... False
463
+ sample_rate ..................................... 1.0
464
+ save ............................................ None
465
+ save_interval ................................... None
466
+ save_retain_interval ............................ None
467
+ scatter_gather_tensors_in_pipeline .............. True
468
+ seed ............................................ 1234
469
+ seq_length ...................................... 8192
470
+ sequence_parallel ............................... False
471
+ sft ............................................. False
472
+ sft_tokenizer_prompt_format ..................... nemotron-h-aligned
473
+ sgd_momentum .................................... 0.9
474
+ short_seq_prob .................................. 0.1
475
+ skip_train ...................................... False
476
+ skipped_train_samples ........................... 0
477
+ spec ............................................ None
478
+ split ........................................... 100,0,0
479
+ squared_relu .................................... False
480
+ start_weight_decay .............................. 0.1
481
+ straggler_ctrlr_port ............................ 65535
482
+ straggler_minmax_count .......................... 1
483
+ suggested_communication_unit_size ............... None
484
+ swiglu .......................................... True
485
+ swin_backbone_type .............................. tiny
486
+ symmetric_ar_type ............................... None
487
+ te_rng_tracker .................................. True
488
+ tensor_model_parallel_size ...................... 1
489
+ tensorboard_dir ................................. None
490
+ tensorboard_log_interval ........................ 1
491
+ tensorboard_queue_size .......................... 1000
492
+ test_data_path .................................. None
493
+ test_mode ....................................... False
494
+ tiktoken_num_special_tokens ..................... 1000
495
+ tiktoken_pattern ................................ None
496
+ tiktoken_special_tokens ......................... None
497
+ timing_log_level ................................ 0
498
+ timing_log_option ............................... minmax
499
+ titles_data_path ................................ None
500
+ tokenizer_model ................................. None
501
+ tokenizer_type .................................. GPT2BPETokenizer
502
+ torch_fsdp2_reshard_after_forward ............... True
503
+ tp_comm_bootstrap_backend ....................... nccl
504
+ tp_comm_bulk_dgrad .............................. True
505
+ tp_comm_bulk_wgrad .............................. True
506
+ tp_comm_overlap ................................. False
507
+ tp_comm_overlap_ag .............................. True
508
+ tp_comm_overlap_cfg ............................. None
509
+ tp_comm_overlap_rs .............................. True
510
+ tp_comm_overlap_rs_dgrad ........................ False
511
+ tp_comm_split_ag ................................ True
512
+ tp_comm_split_rs ................................ True
513
+ train_data_path ................................. None
514
+ train_iters ..................................... 12
515
+ train_samples ................................... None
516
+ train_sync_interval ............................. None
517
+ transformer_impl ................................ transformer_engine
518
+ transformer_pipeline_model_parallel_size ........ 1
519
+ untie_embeddings_and_output_weights ............. True
520
+ use_checkpoint_args ............................. False
521
+ use_checkpoint_opt_param_scheduler .............. False
522
+ use_cpu_initialization .......................... None
523
+ use_custom_fsdp ................................. False
524
+ use_dist_ckpt ................................... True
525
+ use_dist_ckpt_deprecated ........................ False
526
+ use_distributed_optimizer ....................... True
527
+ use_flash_attn .................................. False
528
+ use_legacy_models ............................... False
529
+ use_mp_args_from_checkpoint_args ................ False
530
+ use_one_sent_docs ............................... False
531
+ use_persistent_ckpt_worker ...................... False
532
+ use_precision_aware_optimizer ................... False
533
+ use_pytorch_profiler ............................ False
534
+ use_ring_exchange_p2p ........................... False
535
+ use_rope_scaling ................................ False
536
+ use_rotary_position_embeddings .................. False
537
+ use_sharp ....................................... False
538
+ use_tokenizer_model_from_checkpoint_args ........ True
539
+ use_torch_fsdp2 ................................. False
540
+ use_torch_optimizer_for_cpu_offload ............. False
541
+ use_tp_pp_dp_mapping ............................ False
542
+ v_head_dim ...................................... 128
543
+ valid_data_path ................................. None
544
+ variable_seq_lengths ............................ False
545
+ virtual_pipeline_model_parallel_size ............ None
546
+ vision_backbone_type ............................ vit
547
+ vision_pretraining .............................. False
548
+ vision_pretraining_type ......................... classify
549
+ vocab_extra_ids ................................. 0
550
+ vocab_file ...................................... /workspace/data/wikitext/gpt2-vocab.json
551
+ vocab_size ...................................... None
552
+ wandb_exp_name ..................................
553
+ wandb_project ...................................
554
+ wandb_save_dir ..................................
555
+ weight_decay .................................... 0.1
556
+ weight_decay_incr_style ......................... constant
557
+ wgrad_deferral_limit ............................ 0
558
+ world_size ...................................... 8
559
+ yaml_cfg ........................................ None
560
+ -------------------- end of arguments ---------------------
561
+ INFO:megatron.core.num_microbatches_calculator:setting number of microbatches to constant 2
562
+ > building GPT2BPETokenizer tokenizer ...
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
+ > padded vocab (size: 50257) with 47 dummy tokens (new size: 50304)
568
+ WARNING:megatron.core.rerun_state_machine:RerunStateMachine initialized in mode RerunMode.VALIDATE_RESULTS
569
+ > initializing torch distributed ...
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 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
577
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
578
+ [Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
579
+ [Gloo] Rank 5 is connected to 7 peer ranks. [Gloo] Rank Expected number of connected peer ranks is : 74
580
+ is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
581
+ [Gloo] Rank 2 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
582
+ [Gloo] Rank 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
583
+ [Gloo] Rank 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
584
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
585
+ [Gloo] Rank 0 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 6 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
589
+ [Gloo] Rank [Gloo] Rank [Gloo] Rank 35 is connected to 7 is connected to 7 is connected to 7 peer ranks. 7 peer ranks. Expected number of connected peer ranks is : peer ranks. Expected number of connected peer ranks is : 7Expected number of connected peer ranks is : 7
590
+ 7
591
+
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 [Gloo] Rank 00 is connected to is connected to 00 peer ranks. peer ranks. Expected number of connected peer ranks is : Expected number of connected peer ranks is : 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
+ [rank7]:[W830 06:40:45.890883990 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
+ [rank3]:[W830 06:40:45.891170636 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.
605
+ [rank6]:[W830 06:40:45.891191680 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.
606
+ [rank2]:[W830 06:40:45.891251584 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]:[W830 06:40:45.891258014 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
+ [rank4]:[W830 06:40:45.893069408 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.
609
+ [rank1]:[W830 06:40:45.893093521 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.
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]:[W830 06:40:45.058053452 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: 3.627 seconds
619
+ time to initialize megatron (seconds): 7.321
620
+ [after megatron is initialized] datetime: 2026-08-30 06:40:52
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): 1385828352
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: 18
641
+ Params for bucket 1 (206045184 elements, 206045184 padded size):
642
+ module.output_layer.weight
643
+ Params for bucket 2 (41988096 elements, 41988096 padded size):
644
+ module.decoder.layers.15.mlp.router.weight
645
+ module.decoder.layers.15.self_attention.linear_qkv.layer_norm_weight
646
+ module.decoder.layers.15.self_attention.linear_proj.weight
647
+ module.decoder.final_layernorm.weight
648
+ module.decoder.layers.15.self_attention.linear_qkv.weight
649
+ module.decoder.layers.15.pre_mlp_layernorm.weight
650
+ Params for bucket 3 (41984000 elements, 41984000 padded size):
651
+ module.decoder.layers.14.pre_mlp_layernorm.weight
652
+ module.decoder.layers.14.self_attention.linear_qkv.weight
653
+ module.decoder.layers.14.mlp.router.weight
654
+ module.decoder.layers.14.self_attention.linear_proj.weight
655
+ module.decoder.layers.14.self_attention.linear_qkv.layer_norm_weight
656
+ Params for bucket 4 (41984000 elements, 41984000 padded size):
657
+ module.decoder.layers.13.mlp.router.weight
658
+ module.decoder.layers.13.self_attention.linear_qkv.layer_norm_weight
659
+ module.decoder.layers.13.self_attention.linear_proj.weight
660
+ module.decoder.layers.13.self_attention.linear_qkv.weight
661
+ module.decoder.layers.13.pre_mlp_layernorm.weight
662
+ Params for bucket 5 (41984000 elements, 41984000 padded size):
663
+ module.decoder.layers.12.mlp.router.weight
664
+ module.decoder.layers.12.self_attention.linear_qkv.layer_norm_weight
665
+ module.decoder.layers.12.self_attention.linear_proj.weight
666
+ module.decoder.layers.12.self_attention.linear_qkv.weight
667
+ module.decoder.layers.12.pre_mlp_layernorm.weight
668
+ Params for bucket 6 (41984000 elements, 41984000 padded size):
669
+ module.decoder.layers.11.mlp.router.weight
670
+ module.decoder.layers.11.self_attention.linear_qkv.layer_norm_weight
671
+ module.decoder.layers.11.self_attention.linear_proj.weight
672
+ module.decoder.layers.11.self_attention.linear_qkv.weight
673
+ module.decoder.layers.11.pre_mlp_layernorm.weight
674
+ Params for bucket 7 (41984000 elements, 41984000 padded size):
675
+ module.decoder.layers.10.self_attention.linear_proj.weight
676
+ module.decoder.layers.10.self_attention.linear_qkv.layer_norm_weight
677
+ module.decoder.layers.10.mlp.router.weight
678
+ module.decoder.layers.10.pre_mlp_layernorm.weight
679
+ module.decoder.layers.10.self_attention.linear_qkv.weight
680
+ Params for bucket 8 (41984000 elements, 41984000 padded size):
681
+ module.decoder.layers.9.self_attention.linear_qkv.weight
682
+ module.decoder.layers.9.pre_mlp_layernorm.weight
683
+ module.decoder.layers.9.mlp.router.weight
684
+ module.decoder.layers.9.self_attention.linear_qkv.layer_norm_weight
685
+ module.decoder.layers.9.self_attention.linear_proj.weight
686
+ Params for bucket 9 (41984000 elements, 41984000 padded size):
687
+ module.decoder.layers.8.mlp.router.weight
688
+ module.decoder.layers.8.self_attention.linear_proj.weight
689
+ module.decoder.layers.8.self_attention.linear_qkv.layer_norm_weight
690
+ module.decoder.layers.8.self_attention.linear_qkv.weight
691
+ module.decoder.layers.8.pre_mlp_layernorm.weight
692
+ Params for bucket 10 (41984000 elements, 41984000 padded size):
693
+ module.decoder.layers.7.self_attention.linear_qkv.weight
694
+ module.decoder.layers.7.pre_mlp_layernorm.weight
695
+ module.decoder.layers.7.mlp.router.weight
696
+ module.decoder.layers.7.self_attention.linear_qkv.layer_norm_weight
697
+ module.decoder.layers.7.self_attention.linear_proj.weight
698
+ Params for bucket 11 (41984000 elements, 41984000 padded size):
699
+ module.decoder.layers.6.mlp.router.weight
700
+ module.decoder.layers.6.self_attention.linear_qkv.layer_norm_weight
701
+ module.decoder.layers.6.pre_mlp_layernorm.weight
702
+ module.decoder.layers.6.self_attention.linear_qkv.weight
703
+ module.decoder.layers.6.self_attention.linear_proj.weight
704
+ Params for bucket 12 (41984000 elements, 41984000 padded size):
705
+ module.decoder.layers.5.self_attention.linear_qkv.weight
706
+ module.decoder.layers.5.self_attention.linear_proj.weight
707
+ module.decoder.layers.5.pre_mlp_layernorm.weight
708
+ module.decoder.layers.5.mlp.router.weight
709
+ module.decoder.layers.5.self_attention.linear_qkv.layer_norm_weight
710
+ Params for bucket 13 (41984000 elements, 41984000 padded size):
711
+ module.decoder.layers.4.self_attention.linear_qkv.weight
712
+ module.decoder.layers.4.pre_mlp_layernorm.weight
713
+ module.decoder.layers.4.mlp.router.weight
714
+ module.decoder.layers.4.self_attention.linear_qkv.layer_norm_weight
715
+ module.decoder.layers.4.self_attention.linear_proj.weight
716
+ Params for bucket 14 (41984000 elements, 41984000 padded size):
717
+ module.decoder.layers.3.mlp.router.weight
718
+ module.decoder.layers.3.self_attention.linear_qkv.weight
719
+ module.decoder.layers.3.pre_mlp_layernorm.weight
720
+ module.decoder.layers.3.self_attention.linear_proj.weight
721
+ module.decoder.layers.3.self_attention.linear_qkv.layer_norm_weight
722
+ Params for bucket 15 (41984000 elements, 41984000 padded size):
723
+ module.decoder.layers.2.mlp.router.weight
724
+ module.decoder.layers.2.self_attention.linear_qkv.weight
725
+ module.decoder.layers.2.pre_mlp_layernorm.weight
726
+ module.decoder.layers.2.self_attention.linear_proj.weight
727
+ module.decoder.layers.2.self_attention.linear_qkv.layer_norm_weight
728
+ Params for bucket 16 (41984000 elements, 41984000 padded size):
729
+ module.decoder.layers.1.self_attention.linear_qkv.weight
730
+ module.decoder.layers.1.pre_mlp_layernorm.weight
731
+ module.decoder.layers.1.mlp.router.weight
732
+ module.decoder.layers.1.self_attention.linear_qkv.layer_norm_weight
733
+ module.decoder.layers.1.self_attention.linear_proj.weight
734
+ Params for bucket 17 (41984000 elements, 41984000 padded size):
735
+ module.decoder.layers.0.mlp.router.weight
736
+ module.decoder.layers.0.self_attention.linear_qkv.weight
737
+ module.decoder.layers.0.self_attention.linear_qkv.layer_norm_weight
738
+ module.decoder.layers.0.pre_mlp_layernorm.weight
739
+ module.decoder.layers.0.self_attention.linear_proj.weight
740
+ Params for bucket 18 (206045184 elements, 206045184 padded size):
741
+ module.embedding.word_embeddings.weight
742
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 7
743
+ Params for bucket 1 (44040192 elements, 44040192 padded size):
744
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight0
745
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight0
746
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight0
747
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight0
748
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight0
749
+ Params for bucket 2 (50331648 elements, 50331648 padded size):
750
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight0
751
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight0
752
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight0
753
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight0
754
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight0
755
+ Params for bucket 3 (44040192 elements, 44040192 padded size):
756
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight0
757
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight0
758
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight0
759
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight0
760
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight0
761
+ Params for bucket 4 (50331648 elements, 50331648 padded size):
762
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight0
763
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight0
764
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight0
765
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight0
766
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight0
767
+ Params for bucket 5 (44040192 elements, 44040192 padded size):
768
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight0
769
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight0
770
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight0
771
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight0
772
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight0
773
+ Params for bucket 6 (50331648 elements, 50331648 padded size):
774
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight0
775
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight0
776
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight0
777
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight0
778
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight0
779
+ Params for bucket 7 (18874368 elements, 18874368 padded size):
780
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight0
781
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight0
782
+ 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 0x7473683a85c0>, config_logger_dir='')
783
+ INFO:megatron.core.optimizer_param_scheduler:> learning rate decay style: cosine
784
+ [after model, optimizer, and learning rate scheduler are built] datetime: 2026-08-30 06:40:52
785
+ > building train, validation, and test datasets ...
786
+ > datasets target sizes (minimum size):
787
+ train: 192
788
+ validation: 0
789
+ test: 0
790
+ INFO:megatron.core.datasets.blended_megatron_dataset_config:Let split_matrix = [(0, 1.0), None, None]
791
+ > building train, validation, and test datasets for GPT ...
792
+ INFO:megatron.core.datasets.blended_megatron_dataset_builder:Building GPTDataset splits with sizes=(192, 0, 0) and config=GPTDatasetConfig(random_seed=1234, sequence_length=8192, 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 0x74736a5026c0>, 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)
793
+ INFO:megatron.core.datasets.indexed_dataset:Load the _IndexReader from /workspace/data/wikitext/wikitext103_gpt2_text_document.idx
794
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence lengths
795
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence pointers
796
+ INFO:megatron.core.datasets.indexed_dataset: Extract the document indices
797
+ INFO:megatron.core.datasets.indexed_dataset:> total number of sequences: 1165029
798
+ INFO:megatron.core.datasets.indexed_dataset:> total number of documents: 1801350
799
+ INFO:megatron.core.datasets.gpt_dataset:Build and save the GPTDataset train indices
800
+ INFO:megatron.core.datasets.gpt_dataset:> total number of samples: 14536
801
+ INFO:megatron.core.datasets.gpt_dataset:> total number of epochs: 1
802
+ > finished creating GPT datasets ...
803
+ [after dataloaders are built] datetime: 2026-08-30 06:40:52
804
+ done with setup ...
805
+ (min, max) time across ranks (ms):
806
+ model-and-optimizer-setup ......................: (305.66, 305.91)
807
+ train/valid/test-data-iterators-setup ..........: (123.80, 134.93)
808
+ training ...
809
+ Overwriting rerun_state_machine.current_iteration from -1 to 0...
810
+ [before the start of training step] datetime: 2026-08-30 06:40:52
811
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
812
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
813
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
814
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
815
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
816
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
817
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
818
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
819
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
820
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
821
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
822
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
823
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
824
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
825
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
826
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
827
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
828
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
829
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
830
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
831
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
832
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
833
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
834
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
835
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
836
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
837
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
838
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
839
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
840
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
841
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
842
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
843
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
844
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
845
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
846
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
847
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
848
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
849
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
850
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
851
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
852
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
853
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
854
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
855
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
856
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
857
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
858
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
859
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
860
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
861
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
862
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
863
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
864
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
865
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
866
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
867
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
868
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
869
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
870
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
871
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
872
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
873
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
874
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
875
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
876
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
877
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
878
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
879
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
880
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
881
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
882
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
883
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
884
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
885
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
886
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
887
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
888
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
889
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
890
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
891
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
892
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
893
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
894
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
895
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
896
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
897
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
898
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
899
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
900
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
901
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
902
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
903
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
904
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
905
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
906
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
907
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
908
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
909
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
910
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
911
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
912
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
913
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
914
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
915
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
916
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
917
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
918
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
919
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
920
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
921
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
922
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
923
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
924
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
925
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
926
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
927
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
928
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
929
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
930
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
931
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
932
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
933
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
934
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
935
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
936
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
937
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
938
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
939
+ [2026-08-30 06:41:31] iteration 1/ 12 | consumed samples: 16 | elapsed time per iteration (ms): 39131.9 | throughput per GPU (TFLOP/s/GPU): 5.1 | learning rate: 2.000000E-05 | global batch size: 16 | lm loss: 1.103051E+01 | loss scale: 1.0 | grad norm: 49.673 | number of skipped iterations: 0 | number of nan iterations: 0 |
940
+ Number of parameters in transformer block in billions: 3.09
941
+ Number of parameters in embedding layers in billions: 0.41
942
+ Total number of parameters in billions: 3.50
943
+ Number of parameters in most loaded shard in billions: 3.4994
944
+ Theoretical memory footprints: weight and optimizer=25029.43 MB
945
+ [Rank 0] (after 1 iterations) memory (MB) | allocated: 15544.85546875 | max allocated: 31627.193359375 | reserved: 32528.0 | max reserved: 32528.0
946
+ [2026-08-30 06:41:32] iteration 2/ 12 | consumed samples: 32 | elapsed time per iteration (ms): 421.4 | throughput per GPU (TFLOP/s/GPU): 470.7 | learning rate: 4.000000E-05 | global batch size: 16 | lm loss: 1.101006E+01 | loss scale: 1.0 | grad norm: 50.588 | number of skipped iterations: 0 | number of nan iterations: 0 |
947
+ [2026-08-30 06:41:32] iteration 3/ 12 | consumed samples: 48 | elapsed time per iteration (ms): 410.7 | throughput per GPU (TFLOP/s/GPU): 483.0 | learning rate: 6.000000E-05 | global batch size: 16 | lm loss: 1.060387E+01 | loss scale: 1.0 | grad norm: 251.419 | number of skipped iterations: 0 | number of nan iterations: 0 |
948
+ Capture CUDA graph for training!!!
949
+ CUDA graph capture done!!!
950
+ [2026-08-30 06:41:33] iteration 4/ 12 | consumed samples: 64 | elapsed time per iteration (ms): 1271.1 | throughput per GPU (TFLOP/s/GPU): 156.1 | learning rate: 8.000000E-05 | global batch size: 16 | lm loss: 1.310918E+01 | loss scale: 1.0 | grad norm: 40.109 | number of skipped iterations: 0 | number of nan iterations: 0 |
951
+ [2026-08-30 06:41:34] iteration 5/ 12 | consumed samples: 80 | elapsed time per iteration (ms): 316.7 | throughput per GPU (TFLOP/s/GPU): 626.4 | learning rate: 1.000000E-04 | global batch size: 16 | lm loss: 1.053840E+01 | loss scale: 1.0 | grad norm: 19.465 | number of skipped iterations: 0 | number of nan iterations: 0 |
952
+ [2026-08-30 06:41:34] iteration 6/ 12 | consumed samples: 96 | elapsed time per iteration (ms): 317.7 | throughput per GPU (TFLOP/s/GPU): 624.4 | learning rate: 7.750000E-05 | global batch size: 16 | lm loss: 1.033997E+01 | loss scale: 1.0 | grad norm: 8.787 | number of skipped iterations: 0 | number of nan iterations: 0 |
953
+ [2026-08-30 06:41:34] iteration 7/ 12 | consumed samples: 112 | elapsed time per iteration (ms): 315.9 | throughput per GPU (TFLOP/s/GPU): 628.0 | learning rate: 3.250000E-05 | global batch size: 16 | lm loss: 9.519092E+00 | loss scale: 1.0 | grad norm: 7.481 | number of skipped iterations: 0 | number of nan iterations: 0 |
954
+ [2026-08-30 06:41:35] iteration 8/ 12 | consumed samples: 128 | elapsed time per iteration (ms): 316.1 | throughput per GPU (TFLOP/s/GPU): 627.7 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.806551E+00 | loss scale: 1.0 | grad norm: 6.197 | number of skipped iterations: 0 | number of nan iterations: 0 |
955
+ [2026-08-30 06:41:35] iteration 9/ 12 | consumed samples: 144 | elapsed time per iteration (ms): 317.1 | throughput per GPU (TFLOP/s/GPU): 625.5 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.511405E+00 | loss scale: 1.0 | grad norm: 5.371 | number of skipped iterations: 0 | number of nan iterations: 0 |
956
+ [2026-08-30 06:41:35] iteration 10/ 12 | consumed samples: 160 | elapsed time per iteration (ms): 316.2 | throughput per GPU (TFLOP/s/GPU): 627.3 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.421611E+00 | loss scale: 1.0 | grad norm: 4.865 | number of skipped iterations: 0 | number of nan iterations: 0 |
957
+ [2026-08-30 06:41:36] iteration 11/ 12 | consumed samples: 176 | elapsed time per iteration (ms): 315.5 | throughput per GPU (TFLOP/s/GPU): 628.8 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.354515E+00 | loss scale: 1.0 | grad norm: 4.850 | number of skipped iterations: 0 | number of nan iterations: 0 |
958
+ [2026-08-30 06:41:36] iteration 12/ 12 | consumed samples: 192 | elapsed time per iteration (ms): 316.0 | throughput per GPU (TFLOP/s/GPU): 627.7 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.256328E+00 | loss scale: 1.0 | grad norm: 4.447 | number of skipped iterations: 0 | number of nan iterations: 0 |
959
+ [after training is done] datetime: 2026-08-30 06:41:36
960
+ W0830 06:41:36.685000 640669 torch/distributed/elastic/agent/server/api.py:719] Received 15 death signal, shutting down workers
961
+ W0830 06:41:36.686000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640758 closing signal SIGTERM
962
+ W0830 06:41:36.686000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640759 closing signal SIGTERM
963
+ W0830 06:41:36.687000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640760 closing signal SIGTERM
964
+ W0830 06:41:36.691000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640761 closing signal SIGTERM
965
+ W0830 06:41:36.692000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640762 closing signal SIGTERM
966
+ W0830 06:41:36.694000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640763 closing signal SIGTERM
967
+ W0830 06:41:36.695000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640764 closing signal SIGTERM
968
+ W0830 06:41:36.697000 640669 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 640765 closing signal SIGTERM
969
+ Traceback (most recent call last):
970
+ File "/usr/local/bin/torchrun", line 8, in <module>
971
+ sys.exit(main())
972
+ ^^^^^^
973
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper
974
+ return f(*args, **kwargs)
975
+ ^^^^^^^^^^^^^^^^^^
976
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 892, in main
977
+ run(args)
978
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 883, in run
979
+ elastic_launch(
980
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 139, in __call__
981
+ return launch_agent(self._config, self._entrypoint, list(args))
982
+ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
983
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 261, in launch_agent
984
+ result = agent.run()
985
+ ^^^^^^^^^^^
986
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper
987
+ result = f(*args, **kwargs)
988
+ ^^^^^^^^^^^^^^^^^^
989
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run
990
+ result = self._invoke_run(role)
991
+ ^^^^^^^^^^^^^^^^^^^^^^
992
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run
993
+ time.sleep(monitor_interval)
994
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler
995
+ raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
996
+ torch.distributed.elastic.multiprocessing.api.SignalException: Process 640669 got signal: 15
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep2/mem_samples.csv ADDED
@@ -0,0 +1,232 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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, 1
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, 45
43
+ 2, 39
44
+ 3, 629
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, 2007
58
+ 1, 2007
59
+ 2, 2007
60
+ 3, 2007
61
+ 4, 2007
62
+ 5, 2007
63
+ 6, 2007
64
+ 7, 2007
65
+ 0, 12405
66
+ 1, 12401
67
+ 2, 12401
68
+ 3, 12403
69
+ 4, 12403
70
+ 5, 12403
71
+ 6, 12467
72
+ 7, 12403
73
+ 0, 13053
74
+ 1, 13045
75
+ 2, 13047
76
+ 3, 13045
77
+ 4, 13045
78
+ 5, 13059
79
+ 6, 13061
80
+ 7, 13071
81
+ 0, 13813
82
+ 1, 13811
83
+ 2, 13811
84
+ 3, 13811
85
+ 4, 13811
86
+ 5, 13811
87
+ 6, 13811
88
+ 7, 13811
89
+ 0, 13837
90
+ 1, 13835
91
+ 2, 13835
92
+ 3, 13835
93
+ 4, 13835
94
+ 5, 13835
95
+ 6, 13835
96
+ 7, 13835
97
+ 0, 13837
98
+ 1, 13835
99
+ 2, 13835
100
+ 3, 13835
101
+ 4, 13835
102
+ 5, 13835
103
+ 6, 13835
104
+ 7, 13835
105
+ 0, 15009
106
+ 1, 15007
107
+ 2, 15007
108
+ 3, 15007
109
+ 4, 15007
110
+ 5, 15007
111
+ 6, 15007
112
+ 7, 15007
113
+ 0, 15317
114
+ 1, 15315
115
+ 2, 15315
116
+ 3, 15315
117
+ 4, 15315
118
+ 5, 15315
119
+ 6, 15315
120
+ 7, 15315
121
+ 0, 15859
122
+ 1, 15857
123
+ 2, 15857
124
+ 3, 15857
125
+ 4, 15857
126
+ 5, 15857
127
+ 6, 15857
128
+ 7, 15857
129
+ 0, 16283
130
+ 1, 16281
131
+ 2, 16281
132
+ 3, 16281
133
+ 4, 16281
134
+ 5, 16281
135
+ 6, 16281
136
+ 7, 16281
137
+ 0, 34589
138
+ 1, 34587
139
+ 2, 34587
140
+ 3, 34587
141
+ 4, 34549
142
+ 5, 34587
143
+ 6, 34587
144
+ 7, 34587
145
+ 0, 34589
146
+ 1, 34587
147
+ 2, 34587
148
+ 3, 34587
149
+ 4, 34549
150
+ 5, 34587
151
+ 6, 34587
152
+ 7, 34587
153
+ 0, 34589
154
+ 1, 34587
155
+ 2, 34587
156
+ 3, 34587
157
+ 4, 34549
158
+ 5, 34587
159
+ 6, 34587
160
+ 7, 34587
161
+ 0, 34589
162
+ 1, 34587
163
+ 2, 34587
164
+ 3, 34587
165
+ 4, 34549
166
+ 5, 34587
167
+ 6, 34587
168
+ 7, 34587
169
+ 0, 37491
170
+ 1, 37489
171
+ 2, 37489
172
+ 3, 37489
173
+ 4, 37451
174
+ 5, 37489
175
+ 6, 37489
176
+ 7, 37489
177
+ 0, 37879
178
+ 1, 37877
179
+ 2, 37877
180
+ 3, 37877
181
+ 4, 37839
182
+ 5, 37877
183
+ 6, 37877
184
+ 7, 37877
185
+ 0, 38527
186
+ 1, 38525
187
+ 2, 38525
188
+ 3, 38525
189
+ 4, 38487
190
+ 5, 38525
191
+ 6, 38525
192
+ 7, 38525
193
+ 0, 39167
194
+ 1, 39135
195
+ 2, 39181
196
+ 3, 39187
197
+ 4, 39153
198
+ 5, 39215
199
+ 6, 39195
200
+ 7, 39213
201
+ 0, 53093
202
+ 1, 53089
203
+ 2, 53091
204
+ 3, 53089
205
+ 4, 53283
206
+ 5, 53089
207
+ 6, 53089
208
+ 7, 53089
209
+ 0, 53093
210
+ 1, 53089
211
+ 2, 53091
212
+ 3, 53089
213
+ 4, 53283
214
+ 5, 53089
215
+ 6, 53089
216
+ 7, 53089
217
+ 0, 51921
218
+ 1, 51917
219
+ 2, 51919
220
+ 3, 51981
221
+ 4, 52757
222
+ 5, 52237
223
+ 6, 51917
224
+ 7, 51917
225
+ 0, 199
226
+ 1, 458
227
+ 2, 51919
228
+ 3, 1482
229
+ 4, 202
230
+ 5, 202
231
+ 6, 202
232
+ 7, 202
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep2/run.env ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ backend=echo
2
+ config=PONLY
3
+ family=tierA
4
+ repo=/workspace/Megatron-MoE-Echo
5
+ branch=bracing-regions-through-kernels
6
+ commit=9a236793
7
+ num_layers=16
8
+ hidden=4096
9
+ ffn_hidden=14336
10
+ moe_ffn_hidden=1536
11
+ num_experts=8
12
+ topk=2
13
+ experts_per_rank=1
14
+ expert_params=18874368
15
+ heads=32
16
+ kv_groups=8
17
+ kv_channels=
18
+ qk_layernorm=0
19
+ shared_ffn=0
20
+ shared_expert_overlap=0
21
+ moe_token_dispatcher=flex
22
+ mla=0
23
+ dense_first_layers=0
24
+ moe_layers=16
25
+ rotary_base=1000000
26
+ rotary_scaling=
27
+ qkv_bias=0
28
+ router_balancing=aux_loss
29
+ aux_loss_coeff=1e-2
30
+ router_pre_softmax=0
31
+ router_fusion=0
32
+ router_dtype=fp32
33
+ device_initiated_grouped_gemm=1
34
+ norm_epsilon=1e-5
35
+ seq_length=8192
36
+ micro_batch_size=1
37
+ global_batch_size=16
38
+ num_microbatches=2
39
+ assignments_per_rank=16384
40
+ mean_tokens_per_expert=16384
41
+ slot_policy=one
42
+ slots_per_rank=1
43
+ capacity_mode=static
44
+ capacity=2.40625
45
+ cuda_graph=1
46
+ probe_mode=0
47
+ async_grad_reduce=0
48
+ branch_role=timing
49
+ memory_dump_mode=none
50
+ memory_snapshot_path=
51
+ cuda_graph_warmup_steps=3
52
+ recompute=none
53
+ force_balance=0
54
+ train_iters=12
55
+ ep_size=8
56
+ echo_dispatch_overlap=1
57
+ echo_dispatch_recompute=0
58
+ ultraep_quota_min_tokens=
59
+ ultraep_weight_sync_join=
60
+ profile_enable=0
61
+ kernel_trace=0
62
+ fp8_mode=mxfp8
63
+ fp8_recipe=mxfp8
64
+ training_seed=1234
65
+ learning_rate=1e-4
66
+ min_learning_rate=1.0e-5
67
+ lr_warmup_iters=5
68
+ state_dump_enable=0
69
+ state_dump_dir=
70
+ moonep_assert_no_pending_reduce=0
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep2/train.log ADDED
@@ -0,0 +1,999 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==============================================================
2
+ PONLY / echo (slots/rank=1 policy=one)
3
+ E=8 K=2 H'=1536 m=1
4
+ L=16 seq=8192 gbs=16 (2 microbatches)
5
+ cap_mode=static cap=2.40625 cg=1 recompute=none
6
+ memory_dump=none snapshots=/workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep2/memory_snapshots/snapshot.pickle
7
+ repo=/workspace/Megatron-MoE-Echo @ 9a236793
8
+ ==============================================================
9
+ W0830 06:41:43.211000 644795 torch/distributed/run.py:766]
10
+ W0830 06:41:43.211000 644795 torch/distributed/run.py:766] *****************************************
11
+ W0830 06:41:43.211000 644795 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
+ W0830 06:41:43.211000 644795 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 .............................. 8192
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 ................................. 14336
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 ............................... 16
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 ..................................... 4096
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 ................................. 0.01
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 ......................... 32768
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 ............................. 1536
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 ............................ 8
323
+ moe_pad_expert_input_to_capacity ................ False
324
+ moe_per_layer_logging ........................... False
325
+ moe_permute_fusion .............................. False
326
+ moe_received_token_capacity ..................... 2.40625
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 .......................... False
336
+ moe_router_score_function ....................... softmax
337
+ moe_router_topk ................................. 2
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 ............................. 32
371
+ num_channels .................................... 3
372
+ num_classes ..................................... 1000
373
+ num_dataset_builder_threads ..................... 1
374
+ num_distributed_optimizer_instances ............. 1
375
+ num_experts ..................................... 8
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 ................................ 8
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 ..................................... 1000000
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 ...................................... 8192
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 2
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
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
568
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
569
+ 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
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 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
577
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
578
+ [Gloo] Rank 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
579
+ [Gloo] Rank 5 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
580
+ [Gloo] Rank 2 is connected to [Gloo] Rank 7 peer ranks. 4Expected number of connected peer ranks is : is connected to 77 peer ranks.
581
+ Expected number of connected peer ranks is : 7
582
+ [Gloo] Rank 6 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
583
+ [Gloo] Rank 7 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 [Gloo] Rank [Gloo] Rank [Gloo] Rank [Gloo] Rank 124[Gloo] Rank [Gloo] Rank 3 is connected to 5 is connected to is connected to 76 is connected to 7 is connected to 77 is connected to is connected to 7 peer ranks. 7 peer ranks. peer ranks. 77 peer ranks. Expected number of connected peer ranks is : peer ranks. Expected number of connected peer ranks is : Expected number of connected peer ranks is : peer ranks. peer ranks. Expected number of connected peer ranks is : 7Expected number of connected peer ranks is : 77Expected number of connected peer ranks is : Expected number of connected peer ranks is : 7
586
+ 7
587
+ 77
588
+
589
+
590
+
591
+
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 [Gloo] Rank 0[Gloo] Rank 0 is connected to 0 is connected to 0 is connected to 0 peer ranks. 0 peer ranks. Expected number of connected peer ranks is : peer ranks. Expected number of connected peer ranks is : 0Expected number of connected peer ranks is : 0[Gloo] Rank
597
+ 0
598
+
599
+ 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
600
+ [Gloo] Rank [Gloo] Rank 00 is connected to is connected to 00 peer ranks. peer ranks. Expected number of connected peer ranks is : Expected number of connected peer ranks is : 00
601
+
602
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
603
+ [rank5]:[W830 06:42:01.473073975 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.
604
+ [rank2]:[W830 06:42:01.473102254 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.
605
+ [rank6]:[W830 06:42:01.473109350 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.
606
+ [rank7]:[W830 06:42:01.473135753 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.
607
+ [rank4]:[W830 06:42:01.473141857 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.
608
+ [rank3]:[W830 06:42:01.473198874 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.
609
+ [rank1]:[W830 06:42:01.473243383 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.
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: 0.036 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]:[W830 06:42:01.659438885 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: 3.836 seconds
619
+ time to initialize megatron (seconds): 7.393
620
+ [after megatron is initialized] datetime: 2026-08-30 06:42:07
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): 1385828352
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: 18
641
+ Params for bucket 1 (206045184 elements, 206045184 padded size):
642
+ module.output_layer.weight
643
+ Params for bucket 2 (41988096 elements, 41988096 padded size):
644
+ module.decoder.final_layernorm.weight
645
+ module.decoder.layers.15.self_attention.linear_qkv.weight
646
+ module.decoder.layers.15.pre_mlp_layernorm.weight
647
+ module.decoder.layers.15.mlp.router.weight
648
+ module.decoder.layers.15.self_attention.linear_qkv.layer_norm_weight
649
+ module.decoder.layers.15.self_attention.linear_proj.weight
650
+ Params for bucket 3 (41984000 elements, 41984000 padded size):
651
+ module.decoder.layers.14.mlp.router.weight
652
+ module.decoder.layers.14.self_attention.linear_proj.weight
653
+ module.decoder.layers.14.self_attention.linear_qkv.layer_norm_weight
654
+ module.decoder.layers.14.pre_mlp_layernorm.weight
655
+ module.decoder.layers.14.self_attention.linear_qkv.weight
656
+ Params for bucket 4 (41984000 elements, 41984000 padded size):
657
+ module.decoder.layers.13.self_attention.linear_qkv.weight
658
+ module.decoder.layers.13.pre_mlp_layernorm.weight
659
+ module.decoder.layers.13.mlp.router.weight
660
+ module.decoder.layers.13.self_attention.linear_qkv.layer_norm_weight
661
+ module.decoder.layers.13.self_attention.linear_proj.weight
662
+ Params for bucket 5 (41984000 elements, 41984000 padded size):
663
+ module.decoder.layers.12.self_attention.linear_qkv.weight
664
+ module.decoder.layers.12.pre_mlp_layernorm.weight
665
+ module.decoder.layers.12.mlp.router.weight
666
+ module.decoder.layers.12.self_attention.linear_proj.weight
667
+ module.decoder.layers.12.self_attention.linear_qkv.layer_norm_weight
668
+ Params for bucket 6 (41984000 elements, 41984000 padded size):
669
+ module.decoder.layers.11.mlp.router.weight
670
+ module.decoder.layers.11.self_attention.linear_qkv.layer_norm_weight
671
+ module.decoder.layers.11.self_attention.linear_proj.weight
672
+ module.decoder.layers.11.self_attention.linear_qkv.weight
673
+ module.decoder.layers.11.pre_mlp_layernorm.weight
674
+ Params for bucket 7 (41984000 elements, 41984000 padded size):
675
+ module.decoder.layers.10.self_attention.linear_qkv.layer_norm_weight
676
+ module.decoder.layers.10.self_attention.linear_proj.weight
677
+ module.decoder.layers.10.mlp.router.weight
678
+ module.decoder.layers.10.self_attention.linear_qkv.weight
679
+ module.decoder.layers.10.pre_mlp_layernorm.weight
680
+ Params for bucket 8 (41984000 elements, 41984000 padded size):
681
+ module.decoder.layers.9.mlp.router.weight
682
+ module.decoder.layers.9.self_attention.linear_qkv.layer_norm_weight
683
+ module.decoder.layers.9.self_attention.linear_proj.weight
684
+ module.decoder.layers.9.self_attention.linear_qkv.weight
685
+ module.decoder.layers.9.pre_mlp_layernorm.weight
686
+ Params for bucket 9 (41984000 elements, 41984000 padded size):
687
+ module.decoder.layers.8.pre_mlp_layernorm.weight
688
+ module.decoder.layers.8.self_attention.linear_qkv.weight
689
+ module.decoder.layers.8.mlp.router.weight
690
+ module.decoder.layers.8.self_attention.linear_proj.weight
691
+ module.decoder.layers.8.self_attention.linear_qkv.layer_norm_weight
692
+ Params for bucket 10 (41984000 elements, 41984000 padded size):
693
+ module.decoder.layers.7.pre_mlp_layernorm.weight
694
+ module.decoder.layers.7.self_attention.linear_qkv.weight
695
+ module.decoder.layers.7.mlp.router.weight
696
+ module.decoder.layers.7.self_attention.linear_proj.weight
697
+ module.decoder.layers.7.self_attention.linear_qkv.layer_norm_weight
698
+ Params for bucket 11 (41984000 elements, 41984000 padded size):
699
+ module.decoder.layers.6.self_attention.linear_proj.weight
700
+ module.decoder.layers.6.self_attention.linear_qkv.layer_norm_weight
701
+ module.decoder.layers.6.mlp.router.weight
702
+ module.decoder.layers.6.self_attention.linear_qkv.weight
703
+ module.decoder.layers.6.pre_mlp_layernorm.weight
704
+ Params for bucket 12 (41984000 elements, 41984000 padded size):
705
+ module.decoder.layers.5.mlp.router.weight
706
+ module.decoder.layers.5.self_attention.linear_qkv.layer_norm_weight
707
+ module.decoder.layers.5.pre_mlp_layernorm.weight
708
+ module.decoder.layers.5.self_attention.linear_proj.weight
709
+ module.decoder.layers.5.self_attention.linear_qkv.weight
710
+ Params for bucket 13 (41984000 elements, 41984000 padded size):
711
+ module.decoder.layers.4.mlp.router.weight
712
+ module.decoder.layers.4.self_attention.linear_proj.weight
713
+ module.decoder.layers.4.self_attention.linear_qkv.layer_norm_weight
714
+ module.decoder.layers.4.pre_mlp_layernorm.weight
715
+ module.decoder.layers.4.self_attention.linear_qkv.weight
716
+ Params for bucket 14 (41984000 elements, 41984000 padded size):
717
+ module.decoder.layers.3.self_attention.linear_qkv.weight
718
+ module.decoder.layers.3.pre_mlp_layernorm.weight
719
+ module.decoder.layers.3.mlp.router.weight
720
+ module.decoder.layers.3.self_attention.linear_qkv.layer_norm_weight
721
+ module.decoder.layers.3.self_attention.linear_proj.weight
722
+ Params for bucket 15 (41984000 elements, 41984000 padded size):
723
+ module.decoder.layers.2.mlp.router.weight
724
+ module.decoder.layers.2.self_attention.linear_qkv.layer_norm_weight
725
+ module.decoder.layers.2.self_attention.linear_proj.weight
726
+ module.decoder.layers.2.self_attention.linear_qkv.weight
727
+ module.decoder.layers.2.pre_mlp_layernorm.weight
728
+ Params for bucket 16 (41984000 elements, 41984000 padded size):
729
+ module.decoder.layers.1.mlp.router.weight
730
+ module.decoder.layers.1.self_attention.linear_proj.weight
731
+ module.decoder.layers.1.self_attention.linear_qkv.weight
732
+ module.decoder.layers.1.self_attention.linear_qkv.layer_norm_weight
733
+ module.decoder.layers.1.pre_mlp_layernorm.weight
734
+ Params for bucket 17 (41984000 elements, 41984000 padded size):
735
+ module.decoder.layers.0.pre_mlp_layernorm.weight
736
+ module.decoder.layers.0.self_attention.linear_qkv.weight
737
+ module.decoder.layers.0.mlp.router.weight
738
+ module.decoder.layers.0.self_attention.linear_proj.weight
739
+ module.decoder.layers.0.self_attention.linear_qkv.layer_norm_weight
740
+ Params for bucket 18 (206045184 elements, 206045184 padded size):
741
+ module.embedding.word_embeddings.weight
742
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 7
743
+ Params for bucket 1 (44040192 elements, 44040192 padded size):
744
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight0
745
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight0
746
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight0
747
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight0
748
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight0
749
+ Params for bucket 2 (50331648 elements, 50331648 padded size):
750
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight0
751
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight0
752
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight0
753
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight0
754
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight0
755
+ Params for bucket 3 (44040192 elements, 44040192 padded size):
756
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight0
757
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight0
758
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight0
759
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight0
760
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight0
761
+ Params for bucket 4 (50331648 elements, 50331648 padded size):
762
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight0
763
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight0
764
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight0
765
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight0
766
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight0
767
+ Params for bucket 5 (44040192 elements, 44040192 padded size):
768
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight0
769
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight0
770
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight0
771
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight0
772
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight0
773
+ Params for bucket 6 (50331648 elements, 50331648 padded size):
774
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight0
775
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight0
776
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight0
777
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight0
778
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight0
779
+ Params for bucket 7 (18874368 elements, 18874368 padded size):
780
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight0
781
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight0
782
+ 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 0x71fad0380650>, config_logger_dir='')
783
+ INFO:megatron.core.optimizer_param_scheduler:> learning rate decay style: cosine
784
+ [after model, optimizer, and learning rate scheduler are built] datetime: 2026-08-30 06:42:07
785
+ > building train, validation, and test datasets ...
786
+ > datasets target sizes (minimum size):
787
+ train: 192
788
+ validation: 0
789
+ test: 0
790
+ INFO:megatron.core.datasets.blended_megatron_dataset_config:Let split_matrix = [(0, 1.0), None, None]
791
+ > building train, validation, and test datasets for GPT ...
792
+ INFO:megatron.core.datasets.blended_megatron_dataset_builder:Building GPTDataset splits with sizes=(192, 0, 0) and config=GPTDatasetConfig(random_seed=1234, sequence_length=8192, 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 0x71fad08c6690>, 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)
793
+ INFO:megatron.core.datasets.indexed_dataset:Load the _IndexReader from /workspace/data/wikitext/wikitext103_gpt2_text_document.idx
794
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence lengths
795
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence pointers
796
+ INFO:megatron.core.datasets.indexed_dataset: Extract the document indices
797
+ INFO:megatron.core.datasets.indexed_dataset:> total number of sequences: 1165029
798
+ INFO:megatron.core.datasets.indexed_dataset:> total number of documents: 1801350
799
+ INFO:megatron.core.datasets.gpt_dataset:Load the GPTDataset train indices
800
+ INFO:megatron.core.datasets.gpt_dataset: Load the document index from b16a908ae7f33ed67d5ec87f6f633afe-GPTDataset-train-document_index.npy
801
+ INFO:megatron.core.datasets.gpt_dataset: Load the sample index from b16a908ae7f33ed67d5ec87f6f633afe-GPTDataset-train-sample_index.npy
802
+ INFO:megatron.core.datasets.gpt_dataset: Load the shuffle index from b16a908ae7f33ed67d5ec87f6f633afe-GPTDataset-train-shuffle_index.npy
803
+ INFO:megatron.core.datasets.gpt_dataset:> total number of samples: 14536
804
+ > finished creating GPT datasets ...
805
+ [after dataloaders are built] datetime: 2026-08-30 06:42:07
806
+ done with setup ...
807
+ training ...
808
+ Overwriting rerun_state_machine.current_iteration from -1 to 0...
809
+ (min, max) time across ranks (ms):
810
+ model-and-optimizer-setup ......................: (195.29, 195.66)
811
+ train/valid/test-data-iterators-setup ..........: (64.27, 69.06)
812
+ [before the start of training step] datetime: 2026-08-30 06:42:07
813
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
814
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
815
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
816
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
817
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
818
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
819
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
820
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
821
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
822
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
823
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
824
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
825
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
826
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
827
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
828
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
829
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
830
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
831
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
832
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
833
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
834
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
835
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
836
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
837
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
838
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
839
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
840
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
841
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
842
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
843
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
844
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
845
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
846
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
847
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
848
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
849
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
850
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
851
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
852
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
853
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
854
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
855
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
856
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
857
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
858
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
859
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
860
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
861
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
862
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
863
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
864
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
865
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
866
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
867
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
868
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
869
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
870
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
871
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
872
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
873
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
874
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
875
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
876
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
877
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
878
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
879
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
880
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
881
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
882
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
883
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
884
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
885
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
886
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
887
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
888
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
889
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
890
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
891
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
892
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
893
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
894
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
895
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
896
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
897
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
898
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
899
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
900
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
901
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
902
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
903
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
904
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
905
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
906
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
907
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
908
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
909
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
910
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
911
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
912
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
913
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
914
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
915
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
916
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
917
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
918
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
919
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
920
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
921
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
922
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
923
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
924
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
925
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
926
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
927
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
928
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
929
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
930
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
931
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
932
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
933
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
934
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
935
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
936
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
937
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
938
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
939
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
940
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
941
+ [2026-08-30 06:42:45] iteration 1/ 12 | consumed samples: 16 | elapsed time per iteration (ms): 37689.1 | throughput per GPU (TFLOP/s/GPU): 5.3 | learning rate: 2.000000E-05 | global batch size: 16 | lm loss: 1.103051E+01 | loss scale: 1.0 | grad norm: 49.674 | number of skipped iterations: 0 | number of nan iterations: 0 |
942
+ Number of parameters in transformer block in billions: 3.09
943
+ Number of parameters in embedding layers in billions: 0.41
944
+ Total number of parameters in billions: 3.50
945
+ Number of parameters in most loaded shard in billions: 3.4994
946
+ Theoretical memory footprints: weight and optimizer=25029.43 MB
947
+ [Rank 0] (after 1 iterations) memory (MB) | allocated: 15544.990234375 | max allocated: 31626.068359375 | reserved: 32496.0 | max reserved: 32496.0
948
+ [2026-08-30 06:42:45] iteration 2/ 12 | consumed samples: 32 | elapsed time per iteration (ms): 407.1 | throughput per GPU (TFLOP/s/GPU): 487.3 | learning rate: 4.000000E-05 | global batch size: 16 | lm loss: 1.101006E+01 | loss scale: 1.0 | grad norm: 50.587 | number of skipped iterations: 0 | number of nan iterations: 0 |
949
+ [2026-08-30 06:42:46] iteration 3/ 12 | consumed samples: 48 | elapsed time per iteration (ms): 399.1 | throughput per GPU (TFLOP/s/GPU): 497.1 | learning rate: 6.000000E-05 | global batch size: 16 | lm loss: 1.060405E+01 | loss scale: 1.0 | grad norm: 251.419 | number of skipped iterations: 0 | number of nan iterations: 0 |
950
+ Capture CUDA graph for training!!!
951
+ CUDA graph capture done!!!
952
+ [2026-08-30 06:42:47] iteration 4/ 12 | consumed samples: 64 | elapsed time per iteration (ms): 1240.3 | throughput per GPU (TFLOP/s/GPU): 159.9 | learning rate: 8.000000E-05 | global batch size: 16 | lm loss: 1.310965E+01 | loss scale: 1.0 | grad norm: 40.098 | number of skipped iterations: 0 | number of nan iterations: 0 |
953
+ [2026-08-30 06:42:47] iteration 5/ 12 | consumed samples: 80 | elapsed time per iteration (ms): 317.2 | throughput per GPU (TFLOP/s/GPU): 625.4 | learning rate: 1.000000E-04 | global batch size: 16 | lm loss: 1.053845E+01 | loss scale: 1.0 | grad norm: 19.447 | number of skipped iterations: 0 | number of nan iterations: 0 |
954
+ [2026-08-30 06:42:47] iteration 6/ 12 | consumed samples: 96 | elapsed time per iteration (ms): 315.9 | throughput per GPU (TFLOP/s/GPU): 628.0 | learning rate: 7.750000E-05 | global batch size: 16 | lm loss: 1.033706E+01 | loss scale: 1.0 | grad norm: 8.752 | number of skipped iterations: 0 | number of nan iterations: 0 |
955
+ [2026-08-30 06:42:48] iteration 7/ 12 | consumed samples: 112 | elapsed time per iteration (ms): 317.1 | throughput per GPU (TFLOP/s/GPU): 625.6 | learning rate: 3.250000E-05 | global batch size: 16 | lm loss: 9.513631E+00 | loss scale: 1.0 | grad norm: 7.444 | number of skipped iterations: 0 | number of nan iterations: 0 |
956
+ [2026-08-30 06:42:48] iteration 8/ 12 | consumed samples: 128 | elapsed time per iteration (ms): 316.5 | throughput per GPU (TFLOP/s/GPU): 626.8 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.798345E+00 | loss scale: 1.0 | grad norm: 6.202 | number of skipped iterations: 0 | number of nan iterations: 0 |
957
+ [2026-08-30 06:42:48] iteration 9/ 12 | consumed samples: 144 | elapsed time per iteration (ms): 316.0 | throughput per GPU (TFLOP/s/GPU): 627.7 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.506270E+00 | loss scale: 1.0 | grad norm: 5.326 | number of skipped iterations: 0 | number of nan iterations: 0 |
958
+ [2026-08-30 06:42:49] iteration 10/ 12 | consumed samples: 160 | elapsed time per iteration (ms): 316.5 | throughput per GPU (TFLOP/s/GPU): 626.8 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.414983E+00 | loss scale: 1.0 | grad norm: 4.897 | number of skipped iterations: 0 | number of nan iterations: 0 |
959
+ [2026-08-30 06:42:49] iteration 11/ 12 | consumed samples: 176 | elapsed time per iteration (ms): 318.3 | throughput per GPU (TFLOP/s/GPU): 623.2 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.357988E+00 | loss scale: 1.0 | grad norm: 4.790 | number of skipped iterations: 0 | number of nan iterations: 0 |
960
+ [2026-08-30 06:42:49] iteration 12/ 12 | consumed samples: 192 | elapsed time per iteration (ms): 315.8 | throughput per GPU (TFLOP/s/GPU): 628.2 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.263853E+00 | loss scale: 1.0 | grad norm: 4.515 | number of skipped iterations: 0 | number of nan iterations: 0 |
961
+ [after training is done] datetime: 2026-08-30 06:42:49
962
+ [rank0]:[W830 06:42:51.336268915 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())
963
+ W0830 06:42:51.902000 644795 torch/distributed/elastic/agent/server/api.py:719] Received 15 death signal, shutting down workers
964
+ W0830 06:42:51.903000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644884 closing signal SIGTERM
965
+ W0830 06:42:51.903000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644885 closing signal SIGTERM
966
+ W0830 06:42:51.904000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644886 closing signal SIGTERM
967
+ W0830 06:42:51.904000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644887 closing signal SIGTERM
968
+ W0830 06:42:51.905000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644888 closing signal SIGTERM
969
+ W0830 06:42:51.907000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644889 closing signal SIGTERM
970
+ W0830 06:42:51.907000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644890 closing signal SIGTERM
971
+ W0830 06:42:51.908000 644795 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 644891 closing signal SIGTERM
972
+ Traceback (most recent call last):
973
+ File "/usr/local/bin/torchrun", line 8, in <module>
974
+ sys.exit(main())
975
+ ^^^^^^
976
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper
977
+ return f(*args, **kwargs)
978
+ ^^^^^^^^^^^^^^^^^^
979
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 892, in main
980
+ run(args)
981
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 883, in run
982
+ elastic_launch(
983
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 139, in __call__
984
+ return launch_agent(self._config, self._entrypoint, list(args))
985
+ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
986
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 261, in launch_agent
987
+ result = agent.run()
988
+ ^^^^^^^^^^^
989
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper
990
+ result = f(*args, **kwargs)
991
+ ^^^^^^^^^^^^^^^^^^
992
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run
993
+ result = self._invoke_run(role)
994
+ ^^^^^^^^^^^^^^^^^^^^^^
995
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run
996
+ time.sleep(monitor_interval)
997
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler
998
+ raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
999
+ torch.distributed.elastic.multiprocessing.api.SignalException: Process 644795 got signal: 15
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep3/mem_samples.csv ADDED
@@ -0,0 +1,240 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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, 1
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, 1
41
+ 0, 4
42
+ 1, 4
43
+ 2, 4
44
+ 3, 4
45
+ 4, 4
46
+ 5, 4
47
+ 6, 4
48
+ 7, 629
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, 2007
58
+ 1, 2007
59
+ 2, 2007
60
+ 3, 2007
61
+ 4, 2007
62
+ 5, 2007
63
+ 6, 2007
64
+ 7, 2007
65
+ 0, 12403
66
+ 1, 12401
67
+ 2, 12401
68
+ 3, 12401
69
+ 4, 12401
70
+ 5, 12401
71
+ 6, 12401
72
+ 7, 12401
73
+ 0, 12939
74
+ 1, 12937
75
+ 2, 12937
76
+ 3, 12937
77
+ 4, 12937
78
+ 5, 12937
79
+ 6, 12937
80
+ 7, 12937
81
+ 0, 13587
82
+ 1, 13585
83
+ 2, 13585
84
+ 3, 13585
85
+ 4, 13585
86
+ 5, 13585
87
+ 6, 13585
88
+ 7, 13585
89
+ 0, 13813
90
+ 1, 13811
91
+ 2, 13811
92
+ 3, 13811
93
+ 4, 13811
94
+ 5, 13811
95
+ 6, 13811
96
+ 7, 13811
97
+ 0, 13837
98
+ 1, 13835
99
+ 2, 13835
100
+ 3, 13835
101
+ 4, 13835
102
+ 5, 13835
103
+ 6, 13835
104
+ 7, 13835
105
+ 0, 13955
106
+ 1, 13945
107
+ 2, 13975
108
+ 3, 13949
109
+ 4, 13949
110
+ 5, 13931
111
+ 6, 13967
112
+ 7, 13959
113
+ 0, 15009
114
+ 1, 15007
115
+ 2, 15007
116
+ 3, 15007
117
+ 4, 15007
118
+ 5, 15007
119
+ 6, 15007
120
+ 7, 15007
121
+ 0, 15317
122
+ 1, 15315
123
+ 2, 15315
124
+ 3, 15315
125
+ 4, 15315
126
+ 5, 15315
127
+ 6, 15315
128
+ 7, 15315
129
+ 0, 16283
130
+ 1, 16281
131
+ 2, 16281
132
+ 3, 16281
133
+ 4, 16281
134
+ 5, 16281
135
+ 6, 16281
136
+ 7, 16281
137
+ 0, 16283
138
+ 1, 16281
139
+ 2, 16281
140
+ 3, 16281
141
+ 4, 16281
142
+ 5, 16281
143
+ 6, 16281
144
+ 7, 16281
145
+ 0, 34589
146
+ 1, 34587
147
+ 2, 34587
148
+ 3, 34587
149
+ 4, 34549
150
+ 5, 34587
151
+ 6, 34587
152
+ 7, 34587
153
+ 0, 34589
154
+ 1, 34587
155
+ 2, 34587
156
+ 3, 34587
157
+ 4, 34549
158
+ 5, 34587
159
+ 6, 34587
160
+ 7, 34587
161
+ 0, 34589
162
+ 1, 34587
163
+ 2, 34587
164
+ 3, 34587
165
+ 4, 34549
166
+ 5, 34587
167
+ 6, 34587
168
+ 7, 34587
169
+ 0, 34589
170
+ 1, 34587
171
+ 2, 34587
172
+ 3, 34587
173
+ 4, 34549
174
+ 5, 34587
175
+ 6, 34587
176
+ 7, 34587
177
+ 0, 37491
178
+ 1, 37489
179
+ 2, 37489
180
+ 3, 37489
181
+ 4, 37451
182
+ 5, 37489
183
+ 6, 37489
184
+ 7, 37489
185
+ 0, 37975
186
+ 1, 37981
187
+ 2, 37971
188
+ 3, 37971
189
+ 4, 37937
190
+ 5, 37963
191
+ 6, 37973
192
+ 7, 37981
193
+ 0, 38821
194
+ 1, 38805
195
+ 2, 38799
196
+ 3, 38809
197
+ 4, 38797
198
+ 5, 38875
199
+ 6, 38807
200
+ 7, 38843
201
+ 0, 43597
202
+ 1, 43637
203
+ 2, 43637
204
+ 3, 43637
205
+ 4, 43601
206
+ 5, 43637
207
+ 6, 43637
208
+ 7, 43637
209
+ 0, 53093
210
+ 1, 53087
211
+ 2, 53089
212
+ 3, 53089
213
+ 4, 53283
214
+ 5, 53089
215
+ 6, 53089
216
+ 7, 53091
217
+ 0, 53093
218
+ 1, 53087
219
+ 2, 53089
220
+ 3, 53089
221
+ 4, 53283
222
+ 5, 53089
223
+ 6, 53089
224
+ 7, 53091
225
+ 0, 51921
226
+ 1, 51915
227
+ 2, 1738
228
+ 3, 51917
229
+ 4, 52111
230
+ 5, 51917
231
+ 6, 51917
232
+ 7, 51919
233
+ 0, 51918
234
+ 1, 1479
235
+ 2, 199
236
+ 3, 199
237
+ 4, 199
238
+ 5, 199
239
+ 6, 199
240
+ 7, 457
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep3/run.env ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ backend=echo
2
+ config=PONLY
3
+ family=tierA
4
+ repo=/workspace/Megatron-MoE-Echo
5
+ branch=bracing-regions-through-kernels
6
+ commit=9a236793
7
+ num_layers=16
8
+ hidden=4096
9
+ ffn_hidden=14336
10
+ moe_ffn_hidden=1536
11
+ num_experts=8
12
+ topk=2
13
+ experts_per_rank=1
14
+ expert_params=18874368
15
+ heads=32
16
+ kv_groups=8
17
+ kv_channels=
18
+ qk_layernorm=0
19
+ shared_ffn=0
20
+ shared_expert_overlap=0
21
+ moe_token_dispatcher=flex
22
+ mla=0
23
+ dense_first_layers=0
24
+ moe_layers=16
25
+ rotary_base=1000000
26
+ rotary_scaling=
27
+ qkv_bias=0
28
+ router_balancing=aux_loss
29
+ aux_loss_coeff=1e-2
30
+ router_pre_softmax=0
31
+ router_fusion=0
32
+ router_dtype=fp32
33
+ device_initiated_grouped_gemm=1
34
+ norm_epsilon=1e-5
35
+ seq_length=8192
36
+ micro_batch_size=1
37
+ global_batch_size=16
38
+ num_microbatches=2
39
+ assignments_per_rank=16384
40
+ mean_tokens_per_expert=16384
41
+ slot_policy=one
42
+ slots_per_rank=1
43
+ capacity_mode=static
44
+ capacity=2.40625
45
+ cuda_graph=1
46
+ probe_mode=0
47
+ async_grad_reduce=0
48
+ branch_role=timing
49
+ memory_dump_mode=none
50
+ memory_snapshot_path=
51
+ cuda_graph_warmup_steps=3
52
+ recompute=none
53
+ force_balance=0
54
+ train_iters=12
55
+ ep_size=8
56
+ echo_dispatch_overlap=1
57
+ echo_dispatch_recompute=0
58
+ ultraep_quota_min_tokens=
59
+ ultraep_weight_sync_join=
60
+ profile_enable=0
61
+ kernel_trace=0
62
+ fp8_mode=mxfp8
63
+ fp8_recipe=mxfp8
64
+ training_seed=1234
65
+ learning_rate=1e-4
66
+ min_learning_rate=1.0e-5
67
+ lr_warmup_iters=5
68
+ state_dump_enable=0
69
+ state_dump_dir=
70
+ moonep_assert_no_pending_reduce=0
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep3/train.log ADDED
@@ -0,0 +1,999 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ==============================================================
2
+ PONLY / echo (slots/rank=1 policy=one)
3
+ E=8 K=2 H'=1536 m=1
4
+ L=16 seq=8192 gbs=16 (2 microbatches)
5
+ cap_mode=static cap=2.40625 cg=1 recompute=none
6
+ memory_dump=none snapshots=/workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep3/memory_snapshots/snapshot.pickle
7
+ repo=/workspace/Megatron-MoE-Echo @ 9a236793
8
+ ==============================================================
9
+ W0830 06:42:57.406000 648203 torch/distributed/run.py:766]
10
+ W0830 06:42:57.406000 648203 torch/distributed/run.py:766] *****************************************
11
+ W0830 06:42:57.406000 648203 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
+ W0830 06:42:57.406000 648203 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
+ 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
16
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
17
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
18
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
19
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
20
+ 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
21
+ WARNING: overriding default arguments for tokenizer_type:GPT2BPETokenizer with tokenizer_type:GPT2BPETokenizer
22
+ Number of virtual stages per pipeline stage: None
23
+ accumulate and all-reduce gradients in fp32 for bfloat16 data type.
24
+ using torch.bfloat16 for parameters ...
25
+ /workspace/Megatron-MoE-Echo/megatron/training/arguments.py:831: UserWarning: Disabling sequence parallelism because tensor model parallelism is disabled
26
+ warnings.warn("Disabling sequence parallelism because tensor model parallelism is disabled")
27
+ ------------------------ arguments ------------------------
28
+ account_for_embedding_in_pipeline_split ......... False
29
+ account_for_loss_in_pipeline_split .............. False
30
+ accumulate_allreduce_grads_in_fp32 .............. True
31
+ adam_beta1 ...................................... 0.9
32
+ adam_beta2 ...................................... 0.999
33
+ adam_eps ........................................ 1e-08
34
+ add_bias_linear ................................. False
35
+ add_position_embedding .......................... False
36
+ add_qkv_bias .................................... False
37
+ adlr_autoresume ................................. False
38
+ adlr_autoresume_interval ........................ 1000
39
+ align_grad_reduce ............................... True
40
+ align_param_gather .............................. False
41
+ app_tag_run_name ................................ None
42
+ app_tag_run_version ............................. 0.0.0
43
+ apply_layernorm_1p .............................. False
44
+ apply_query_key_layer_scaling ................... False
45
+ apply_residual_connection_post_layernorm ........ False
46
+ apply_rope_fusion ............................... True
47
+ async_save ...................................... None
48
+ async_tensor_model_parallel_allreduce ........... True
49
+ attention_backend ............................... AttnBackend.auto
50
+ attention_dropout ............................... 0.0
51
+ attention_softmax_in_fp32 ....................... False
52
+ auto_detect_ckpt_format ......................... False
53
+ barrier_with_L1_time ............................ True
54
+ bert_binary_head ................................ True
55
+ bert_embedder_type .............................. megatron
56
+ bert_load ....................................... None
57
+ bf16 ............................................ True
58
+ bias_dropout_fusion ............................. True
59
+ bias_gelu_fusion ................................ False
60
+ bias_swiglu_fusion .............................. True
61
+ biencoder_projection_dim ........................ 0
62
+ biencoder_shared_query_context_model ............ False
63
+ block_data_path ................................. None
64
+ calc_ft_timeouts ................................ False
65
+ calculate_per_token_loss ........................ False
66
+ check_for_large_grads ........................... False
67
+ check_for_nan_in_loss_and_grad .................. False
68
+ check_for_spiky_loss ............................ False
69
+ check_weight_hash_across_dp_replicas_interval ... None
70
+ ckpt_assume_constant_structure .................. False
71
+ ckpt_convert_format ............................. None
72
+ ckpt_convert_save ............................... None
73
+ ckpt_convert_update_legacy_dist_opt_format ...... False
74
+ ckpt_format ..................................... torch_dist
75
+ ckpt_fully_parallel_load ........................ False
76
+ ckpt_fully_parallel_save ........................ True
77
+ ckpt_fully_parallel_save_deprecated ............. False
78
+ ckpt_step ....................................... None
79
+ classes_fraction ................................ 1.0
80
+ clip_grad ....................................... 1.0
81
+ clone_scatter_output_in_embedding ............... True
82
+ config_logger_dir ...............................
83
+ consumed_train_samples .......................... 0
84
+ consumed_valid_samples .......................... 0
85
+ context_parallel_size ........................... 1
86
+ cp_comm_type .................................... ['p2p']
87
+ create_attention_mask_in_dataloader ............. True
88
+ cross_entropy_fusion_impl ....................... native
89
+ cross_entropy_loss_fusion ....................... False
90
+ cuda_graph_scope ................................ full_iteration
91
+ cuda_graph_warmup_steps ......................... 3
92
+ data_args_path .................................. None
93
+ data_cache_path ................................. None
94
+ data_parallel_random_init ....................... False
95
+ data_parallel_sharding_strategy ................. no_shard
96
+ data_parallel_size .............................. 8
97
+ data_path ....................................... ['/workspace/data/wikitext/wikitext103_gpt2_text_document']
98
+ data_per_class_fraction ......................... 1.0
99
+ data_sharding ................................... True
100
+ dataloader_type ................................. single
101
+ ddp_average_in_collective ....................... False
102
+ ddp_bucket_size ................................. None
103
+ ddp_num_buckets ................................. None
104
+ ddp_pad_buckets_for_high_nccl_busbw ............. False
105
+ decoder_first_pipeline_num_layers ............... None
106
+ decoder_last_pipeline_num_layers ................ None
107
+ decoder_num_layers .............................. None
108
+ decoder_seq_length .............................. None
109
+ decoupled_lr .................................... None
110
+ decoupled_min_lr ................................ None
111
+ decrease_batch_size_if_needed ................... False
112
+ defer_embedding_wgrad_compute ................... False
113
+ delay_wgrad_compute ............................. False
114
+ deprecated_use_mcore_models ..................... True
115
+ deterministic_mode .............................. False
116
+ dino_bottleneck_size ............................ 256
117
+ dino_freeze_last_layer .......................... 1
118
+ dino_head_hidden_size ........................... 2048
119
+ dino_local_crops_number ......................... 10
120
+ dino_local_img_size ............................. 96
121
+ dino_norm_last_layer ............................ False
122
+ dino_teacher_temp ............................... 0.07
123
+ dino_warmup_teacher_temp ........................ 0.04
124
+ dino_warmup_teacher_temp_epochs ................. 30
125
+ disable_bf16_reduced_precision_matmul ........... False
126
+ disable_mamba_mem_eff_path ...................... False
127
+ disable_straggler_on_startup .................... False
128
+ dist_ckpt_format_deprecated ..................... None
129
+ dist_ckpt_strictness ............................ assume_ok_unexpected
130
+ distribute_saved_activations .................... False
131
+ distributed_backend ............................. nccl
132
+ distributed_timeout_minutes ..................... 10
133
+ embedding_init_method_std ....................... None
134
+ embedding_path .................................. None
135
+ empty_unused_memory_level ....................... 0
136
+ enable_cuda_graph ............................... True
137
+ enable_experimental ............................. False
138
+ enable_ft_package ............................... False
139
+ enable_gloo_process_groups ...................... True
140
+ enable_msc ...................................... True
141
+ enable_one_logger ............................... True
142
+ encoder_num_layers .............................. 16
143
+ encoder_seq_length .............................. 8192
144
+ end_weight_decay ................................ 0.1
145
+ eod_mask_loss ................................... False
146
+ error_injection_rate ............................ 0
147
+ error_injection_type ............................ transient_error
148
+ eval_interval ................................... 1000000
149
+ eval_iters ...................................... 0
150
+ evidence_data_path .............................. None
151
+ exit_duration_in_mins ........................... None
152
+ exit_interval ................................... None
153
+ exit_on_missing_checkpoint ...................... False
154
+ exit_signal_handler ............................. False
155
+ exp_avg_dtype ................................... torch.float32
156
+ exp_avg_sq_dtype ................................ torch.float32
157
+ expert_model_parallel_size ...................... 8
158
+ expert_tensor_parallel_size ..................... 1
159
+ export_force_local_attention .................... False
160
+ export_kd_cfg ................................... None
161
+ export_kd_teacher_ckpt_format ................... None
162
+ export_kd_teacher_load .......................... None
163
+ export_kv_cache_quant ........................... False
164
+ export_legacy_megatron .......................... False
165
+ export_model_type ............................... GPTModel
166
+ export_moe_apply_probs_on_input ................. False
167
+ export_qk_l2_norm ............................... False
168
+ export_quant_cfg ................................ None
169
+ export_real_quant_cfg ........................... None
170
+ export_te_mcore_model ........................... False
171
+ external_cuda_graph ............................. False
172
+ ffn_hidden_size ................................. 14336
173
+ finetune ........................................ False
174
+ finetune_data_split ............................. train
175
+ finetune_hf_dataset ............................. None
176
+ first_last_layers_bf16 .......................... False
177
+ flash_decode .................................... False
178
+ fp16 ............................................ False
179
+ fp16_lm_cross_entropy ........................... False
180
+ fp32_residual_connection ........................ False
181
+ fp8 ............................................. e4m3
182
+ fp8_amax_compute_algo ........................... most_recent
183
+ fp8_amax_history_len ............................ 1
184
+ fp8_interval .................................... 1
185
+ fp8_margin ...................................... 0
186
+ fp8_param_gather ................................ False
187
+ fp8_recipe ...................................... mxfp8
188
+ fp8_wgrad ....................................... True
189
+ fsdp_double_buffer .............................. False
190
+ global_batch_size ............................... 16
191
+ grad_reduce_in_bf16 ............................. False
192
+ gradient_accumulation_fusion .................... True
193
+ gradient_reduce_div_fusion ...................... True
194
+ group_query_attention ........................... True
195
+ head_lr_mult .................................... 1.0
196
+ heterogeneous_layers_config_encoded_json ........ None
197
+ heterogeneous_layers_config_path ................ None
198
+ hidden_dropout .................................. 0.0
199
+ hidden_size ..................................... 4096
200
+ hierarchical_context_parallel_sizes ............. None
201
+ high_priority_stream_groups ..................... []
202
+ hybrid_attention_ratio .......................... 0.0
203
+ hybrid_mlp_ratio ................................ 0.0
204
+ hybrid_override_pattern ......................... None
205
+ hysteresis ...................................... 2
206
+ ict_head_size ................................... None
207
+ ict_load ........................................ None
208
+ img_h ........................................... 224
209
+ img_w ........................................... 224
210
+ indexer_batch_size .............................. 128
211
+ indexer_log_interval ............................ 1000
212
+ inference_batch_times_seqlen_threshold .......... -1
213
+ inference_dynamic_batching ...................... False
214
+ inference_dynamic_batching_buffer_guaranteed_fraction 0.2
215
+ inference_dynamic_batching_buffer_overflow_factor None
216
+ inference_dynamic_batching_buffer_size_gb ....... 40.0
217
+ inference_dynamic_batching_chunk_size ........... 256
218
+ inference_dynamic_batching_max_requests_override None
219
+ inference_dynamic_batching_max_tokens_override .. None
220
+ inference_dynamic_batching_num_cuda_graphs ...... 16
221
+ inference_max_batch_size ........................ 8
222
+ inference_max_seq_length ........................ 2560
223
+ inference_rng_tracker ........................... False
224
+ init_method_std ................................. 0.01
225
+ init_method_xavier_uniform ...................... False
226
+ init_model_with_meta_device ..................... False
227
+ initial_loss_scale .............................. 4294967296
228
+ inprocess_active_world_size ..................... 8
229
+ inprocess_barrier_timeout ....................... 120
230
+ inprocess_completion_timeout .................... 120
231
+ inprocess_empty_cuda_cache ...................... False
232
+ inprocess_granularity ........................... node
233
+ inprocess_hard_timeout .......................... 90
234
+ inprocess_heartbeat_interval .................... 30
235
+ inprocess_heartbeat_timeout ..................... 60
236
+ inprocess_last_call_wait ........................ 1
237
+ inprocess_max_iterations ........................ None
238
+ inprocess_monitor_process_interval .............. 1.0
239
+ inprocess_monitor_thread_interval ............... 1.0
240
+ inprocess_progress_watchdog_interval ............ 1.0
241
+ inprocess_restart ............................... False
242
+ inprocess_soft_timeout .......................... 60
243
+ inprocess_termination_grace_time ................ 1
244
+ is_hybrid_model ................................. False
245
+ iter_per_epoch .................................. 1250
246
+ iterations_to_skip .............................. []
247
+ keep_fp8_transpose_cache_when_using_custom_fsdp . False
248
+ kitchen_config_file ............................. None
249
+ kitchen_recipe_number ........................... None
250
+ kv_channels ..................................... 128
251
+ kv_lora_rank .................................... 32
252
+ lazy_mpu_init ................................... None
253
+ load ............................................ None
254
+ load_main_params_from_ckpt ...................... None
255
+ load_model_opt_format ........................... False
256
+ local_rank ...................................... 0
257
+ log_energy ...................................... False
258
+ log_interval .................................... 1
259
+ log_loss_scale_to_tensorboard ................... True
260
+ log_memory_to_tensorboard ....................... False
261
+ log_num_zeros_in_grad ........................... False
262
+ log_params_norm ................................. False
263
+ log_progress .................................... False
264
+ log_straggler ................................... False
265
+ log_throughput .................................. True
266
+ log_timers_to_tensorboard ....................... False
267
+ log_validation_ppl_to_tensorboard ............... False
268
+ log_world_size_to_tensorboard ................... False
269
+ logging_level ................................... None
270
+ loss_scale ...................................... None
271
+ loss_scale_window ............................... 1000
272
+ lr .............................................. 0.0001
273
+ lr_decay_iters .................................. 8
274
+ lr_decay_samples ................................ None
275
+ lr_decay_style .................................. cosine
276
+ lr_warmup_fraction .............................. None
277
+ lr_warmup_init .................................. 0.0
278
+ lr_warmup_iters ................................. 5
279
+ lr_warmup_samples ............................... 0
280
+ lr_wsd_decay_iters .............................. None
281
+ lr_wsd_decay_samples ............................ None
282
+ lr_wsd_decay_style .............................. exponential
283
+ main_grads_dtype ................................ torch.float32
284
+ main_params_dtype ............................... torch.float32
285
+ make_vocab_size_divisible_by .................... 128
286
+ mamba_head_dim .................................. 64
287
+ mamba_num_groups ................................ 8
288
+ mamba_num_heads ................................. None
289
+ mamba_state_dim ................................. 128
290
+ manual_gc ....................................... False
291
+ manual_gc_eval .................................. True
292
+ manual_gc_interval .............................. 0
293
+ mask_factor ..................................... 1.0
294
+ mask_prob ....................................... 0.15
295
+ mask_type ....................................... random
296
+ masked_softmax_fusion ........................... False
297
+ max_position_embeddings ......................... 32768
298
+ max_tokens_to_oom ............................... 12000
299
+ memory_snapshot_path ............................ snapshot.pickle
300
+ merge_file ...................................... /workspace/data/wikitext/gpt2-merges.txt
301
+ micro_batch_size ................................ 1
302
+ microbatch_group_size_per_vp_stage .............. None
303
+ mid_level_dataset_surplus ....................... 0.005
304
+ min_loss_scale .................................. 1.0
305
+ min_lr .......................................... 1e-05
306
+ mlp_chunks_for_prefill .......................... 1
307
+ mmap_bin_files .................................. True
308
+ mock_data ....................................... False
309
+ moe_apply_probs_on_input ........................ False
310
+ moe_aux_loss_coeff .............................. 0.01
311
+ moe_deepep_num_sms .............................. 20
312
+ moe_echo_enable_random_offloading ............... False
313
+ moe_echo_expert_dispatch_overlap ................ True
314
+ moe_echo_expert_dispatcher_type ................. hybridep
315
+ moe_echo_recompute_expert_dispatch .............. False
316
+ moe_enable_deepep ............................... False
317
+ moe_enable_echo ................................. True
318
+ moe_enable_hybridep ............................. True
319
+ moe_expert_capacity_factor ...................... None
320
+ moe_extended_tp ................................. False
321
+ moe_ffn_hidden_size ............................. 1536
322
+ moe_grouped_gemm ................................ True
323
+ moe_hybridep_num_sms ............................ 16
324
+ moe_input_jitter_eps ............................ None
325
+ moe_layer_freq .................................. 1
326
+ moe_layer_recompute ............................. False
327
+ moe_num_echo_experts ............................ 8
328
+ moe_pad_expert_input_to_capacity ................ False
329
+ moe_per_layer_logging ........................... False
330
+ moe_permute_fusion .............................. False
331
+ moe_received_token_capacity ..................... 2.40625
332
+ moe_router_bias_update_rate ..................... 0.001
333
+ moe_router_dtype ................................ fp32
334
+ moe_router_enable_expert_bias ................... False
335
+ moe_router_force_load_balancing ................. False
336
+ moe_router_group_topk ........................... None
337
+ moe_router_load_balancing_type .................. aux_loss
338
+ moe_router_num_groups ........................... None
339
+ moe_router_padding_for_fp8 ...................... False
340
+ moe_router_pre_softmax .......................... False
341
+ moe_router_score_function ....................... softmax
342
+ moe_router_topk ................................. 2
343
+ moe_router_topk_scaling_factor .................. None
344
+ moe_shared_expert_intermediate_size ............. None
345
+ moe_shared_expert_overlap ....................... False
346
+ moe_token_dispatcher_type ....................... flex
347
+ moe_token_drop_policy ........................... probs
348
+ moe_upcycling_granularity ....................... 1
349
+ moe_use_device_initiated_grouped_gemm ........... True
350
+ moe_use_legacy_grouped_gemm ..................... False
351
+ moe_use_upcycling ............................... False
352
+ moe_z_loss_coeff ................................ None
353
+ mrope_section ................................... None
354
+ mscale .......................................... 1.0
355
+ mscale_all_dim .................................. 0.0
356
+ mtp_loss_scaling_factor ......................... 0.1
357
+ mtp_num_layers .................................. None
358
+ multi_latent_attention .......................... False
359
+ nccl_all_reduce_for_prefill ..................... False
360
+ nccl_communicator_config_path ................... None
361
+ nccl_ub ......................................... False
362
+ no_load_optim ................................... None
363
+ no_load_rng ..................................... None
364
+ no_persist_layer_norm ........................... False
365
+ no_rope_freq .................................... None
366
+ no_save_optim ................................... None
367
+ no_save_rng ..................................... None
368
+ non_persistent_ckpt_type ........................ None
369
+ non_persistent_global_ckpt_dir .................. None
370
+ non_persistent_local_ckpt_algo .................. fully_parallel
371
+ non_persistent_local_ckpt_dir ................... None
372
+ non_persistent_save_interval .................... None
373
+ norm_epsilon .................................... 1e-05
374
+ normalization ................................... RMSNorm
375
+ num_attention_heads ............................. 32
376
+ num_channels .................................... 3
377
+ num_classes ..................................... 1000
378
+ num_dataset_builder_threads ..................... 1
379
+ num_distributed_optimizer_instances ............. 1
380
+ num_experts ..................................... 8
381
+ num_layers ...................................... 16
382
+ num_layers_at_end_in_bf16 ....................... 1
383
+ num_layers_at_start_in_bf16 ..................... 1
384
+ num_layers_per_virtual_pipeline_stage ........... None
385
+ num_query_groups ................................ 8
386
+ num_virtual_stages_per_pipeline_rank ............ None
387
+ num_workers ..................................... 2
388
+ object_storage_cache_path ....................... None
389
+ one_logger_async ................................ False
390
+ one_logger_project .............................. megatron-lm
391
+ one_logger_run_name ............................. None
392
+ onnx_safe ....................................... None
393
+ openai_gelu ..................................... False
394
+ optimizer ....................................... adam
395
+ optimizer_cpu_offload ........................... False
396
+ optimizer_offload_fraction ...................... 1.0
397
+ output_bert_embeddings .......................... False
398
+ overlap_cpu_optimizer_d2h_h2d ................... False
399
+ overlap_grad_reduce ............................. True
400
+ overlap_moe_expert_parallel_comm ................ False
401
+ overlap_p2p_comm ................................ False
402
+ overlap_p2p_comm_warmup_flush ................... False
403
+ overlap_param_gather ............................ True
404
+ overlap_param_gather_with_optimizer_step ........ False
405
+ override_opt_param_scheduler .................... False
406
+ params_dtype .................................... torch.bfloat16
407
+ patch_dim ....................................... 16
408
+ per_split_data_args_path ........................ None
409
+ perform_initialization .......................... True
410
+ pin_cpu_grads ................................... True
411
+ pin_cpu_params .................................. True
412
+ pipeline_model_parallel_comm_backend ............ None
413
+ pipeline_model_parallel_layout .................. None
414
+ pipeline_model_parallel_size .................... 1
415
+ position_embedding_type ......................... rope
416
+ pretrained_checkpoint ........................... None
417
+ profile ......................................... False
418
+ profile_ranks ................................... [0]
419
+ profile_step_end ................................ 12
420
+ profile_step_start .............................. 10
421
+ q_lora_rank ..................................... None
422
+ qk_head_dim ..................................... 128
423
+ qk_l2_norm ...................................... False
424
+ qk_layernorm .................................... False
425
+ qk_pos_emb_head_dim ............................. 64
426
+ query_in_block_prob ............................. 0.1
427
+ rampup_batch_size ............................... None
428
+ rank ............................................ 0
429
+ recompute_granularity ........................... None
430
+ recompute_method ................................ None
431
+ recompute_modules ............................... None
432
+ recompute_num_layers ............................ None
433
+ record_memory_history ........................... False
434
+ relative_attention_max_distance ................. 128
435
+ relative_attention_num_buckets .................. 32
436
+ replication ..................................... False
437
+ replication_factor .............................. 2
438
+ replication_jump ................................ None
439
+ rerun_mode ...................................... validate_results
440
+ reset_attention_mask ............................ False
441
+ reset_position_ids .............................. False
442
+ result_rejected_tracker_filename ................ None
443
+ retriever_report_topk_accuracies ................ []
444
+ retriever_score_scaling ......................... False
445
+ retriever_seq_length ............................ 256
446
+ retro_add_retriever ............................. False
447
+ retro_attention_gate ............................ 1
448
+ retro_cyclic_train_iters ........................ None
449
+ retro_encoder_attention_dropout ................. 0.1
450
+ retro_encoder_hidden_dropout .................... 0.1
451
+ retro_encoder_layers ............................ 2
452
+ retro_num_neighbors ............................. 2
453
+ retro_num_retrieved_chunks ...................... 2
454
+ retro_project_dir ............................... None
455
+ retro_verify_neighbor_count ..................... True
456
+ reuse_grad_buf_for_mxfp8_param_ag ............... False
457
+ rope_scaling_factor ............................. 8.0
458
+ rope_type ....................................... None
459
+ rotary_base ..................................... 1000000
460
+ rotary_interleaved .............................. False
461
+ rotary_percent .................................. 1.0
462
+ rotary_scaling_factor ........................... 1.0
463
+ rotary_seq_len_interpolation_factor ............. None
464
+ run_workload_inspector_server ................... False
465
+ sample_rate ..................................... 1.0
466
+ save ............................................ None
467
+ save_interval ................................... None
468
+ save_retain_interval ............................ None
469
+ scatter_gather_tensors_in_pipeline .............. True
470
+ seed ............................................ 1234
471
+ seq_length ...................................... 8192
472
+ sequence_parallel ............................... False
473
+ sft ............................................. False
474
+ sft_tokenizer_prompt_format ..................... nemotron-h-aligned
475
+ sgd_momentum .................................... 0.9
476
+ short_seq_prob .................................. 0.1
477
+ skip_train ...................................... False
478
+ skipped_train_samples ........................... 0
479
+ spec ............................................ None
480
+ split ........................................... 100,0,0
481
+ squared_relu .................................... False
482
+ start_weight_decay .............................. 0.1
483
+ straggler_ctrlr_port ............................ 65535
484
+ straggler_minmax_count .......................... 1
485
+ suggested_communication_unit_size ............... None
486
+ swiglu .......................................... True
487
+ swin_backbone_type .............................. tiny
488
+ symmetric_ar_type ............................... None
489
+ te_rng_tracker .................................. True
490
+ tensor_model_parallel_size ...................... 1
491
+ tensorboard_dir ................................. None
492
+ tensorboard_log_interval ........................ 1
493
+ tensorboard_queue_size .......................... 1000
494
+ test_data_path .................................. None
495
+ test_mode ....................................... False
496
+ tiktoken_num_special_tokens ..................... 1000
497
+ tiktoken_pattern ................................ None
498
+ tiktoken_special_tokens ......................... None
499
+ timing_log_level ................................ 0
500
+ timing_log_option ............................... minmax
501
+ titles_data_path ................................ None
502
+ tokenizer_model ................................. None
503
+ tokenizer_type .................................. GPT2BPETokenizer
504
+ torch_fsdp2_reshard_after_forward ............... True
505
+ tp_comm_bootstrap_backend ....................... nccl
506
+ tp_comm_bulk_dgrad .............................. True
507
+ tp_comm_bulk_wgrad .............................. True
508
+ tp_comm_overlap ................................. False
509
+ tp_comm_overlap_ag .............................. True
510
+ tp_comm_overlap_cfg ............................. None
511
+ tp_comm_overlap_rs .............................. True
512
+ tp_comm_overlap_rs_dgrad ........................ False
513
+ tp_comm_split_ag ................................ True
514
+ tp_comm_split_rs ................................ True
515
+ train_data_path ................................. None
516
+ train_iters ..................................... 12
517
+ train_samples ................................... None
518
+ train_sync_interval ............................. None
519
+ transformer_impl ................................ transformer_engine
520
+ transformer_pipeline_model_parallel_size ........ 1
521
+ untie_embeddings_and_output_weights ............. True
522
+ use_checkpoint_args ............................. False
523
+ use_checkpoint_opt_param_scheduler .............. False
524
+ use_cpu_initialization .......................... None
525
+ use_custom_fsdp ................................. False
526
+ use_dist_ckpt ................................... True
527
+ use_dist_ckpt_deprecated ........................ False
528
+ use_distributed_optimizer ....................... True
529
+ use_flash_attn .................................. False
530
+ use_legacy_models ............................... False
531
+ use_mp_args_from_checkpoint_args ................ False
532
+ use_one_sent_docs ............................... False
533
+ use_persistent_ckpt_worker ...................... False
534
+ use_precision_aware_optimizer ................... False
535
+ use_pytorch_profiler ............................ False
536
+ use_ring_exchange_p2p ........................... False
537
+ use_rope_scaling ................................ False
538
+ use_rotary_position_embeddings .................. False
539
+ use_sharp ....................................... False
540
+ use_tokenizer_model_from_checkpoint_args ........ True
541
+ use_torch_fsdp2 ................................. False
542
+ use_torch_optimizer_for_cpu_offload ............. False
543
+ use_tp_pp_dp_mapping ............................ False
544
+ v_head_dim ...................................... 128
545
+ valid_data_path ................................. None
546
+ variable_seq_lengths ............................ False
547
+ virtual_pipeline_model_parallel_size ............ None
548
+ vision_backbone_type ............................ vit
549
+ vision_pretraining .............................. False
550
+ vision_pretraining_type ......................... classify
551
+ vocab_extra_ids ................................. 0
552
+ vocab_file ...................................... /workspace/data/wikitext/gpt2-vocab.json
553
+ vocab_size ...................................... None
554
+ wandb_exp_name ..................................
555
+ wandb_project ...................................
556
+ wandb_save_dir ..................................
557
+ weight_decay .................................... 0.1
558
+ weight_decay_incr_style ......................... constant
559
+ wgrad_deferral_limit ............................ 0
560
+ world_size ...................................... 8
561
+ yaml_cfg ........................................ None
562
+ -------------------- end of arguments ---------------------
563
+ INFO:megatron.core.num_microbatches_calculator:setting number of microbatches to constant 2
564
+ > building GPT2BPETokenizer tokenizer ...
565
+ > padded vocab (size: 50257) with 47 dummy tokens (new size: 50304)
566
+ WARNING:megatron.core.rerun_state_machine:RerunStateMachine initialized in mode RerunMode.VALIDATE_RESULTS
567
+ > initializing torch distributed ...
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 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
577
+ [Gloo] Rank 2 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 7 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
580
+ [Gloo] Rank 4 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
581
+ [Gloo] Rank [Gloo] Rank [Gloo] Rank 356 is connected to is connected to is connected to 777 peer ranks. peer ranks. peer ranks. Expected number of connected peer ranks is : Expected number of connected peer ranks is : Expected number of connected peer ranks is : 777
582
+
583
+
584
+ [Gloo] Rank 0 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
585
+ [Gloo] Rank 1 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
586
+ [Gloo] Rank 4 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 3 is connected to 7 peer ranks. Expected number of connected peer ranks is : 7
589
+ [Gloo] Rank 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 [Gloo] Rank 00 is connected to is connected to 00 peer ranks. peer ranks. Expected number of connected peer ranks is : Expected number of connected peer ranks is : 00
597
+
598
+ [Gloo] Rank [Gloo] Rank 0 is connected to 00 is connected to peer ranks. 0Expected number of connected peer ranks is : peer ranks. 0Expected number of connected peer ranks is :
599
+ 0
600
+ [Gloo] Rank [Gloo] Rank 00 is connected to is connected to 00 peer ranks. peer ranks. Expected number of connected peer ranks is : Expected number of connected peer ranks is : 00
601
+
602
+ [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
603
+ [rank2]:[W830 06:43:15.407411422 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.
604
+ [rank1]:[W830 06:43:15.407491665 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
+ [rank6]:[W830 06:43:15.407523630 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.
606
+ [rank7]:[W830 06:43:15.407605947 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.
607
+ [rank3]:[W830 06:43:15.407635755 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
+ [rank4]:[W830 06:43:15.409027036 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.
609
+ [rank5]:[W830 06:43:15.409123934 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.
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]:[W830 06:43:15.564102739 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: 3.730 seconds
619
+ time to initialize megatron (seconds): 7.224
620
+ [after megatron is initialized] datetime: 2026-08-30 06:43:21
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): 1385828352
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: 18
641
+ Params for bucket 1 (206045184 elements, 206045184 padded size):
642
+ module.output_layer.weight
643
+ Params for bucket 2 (41988096 elements, 41988096 padded size):
644
+ module.decoder.layers.15.mlp.router.weight
645
+ module.decoder.layers.15.self_attention.linear_qkv.layer_norm_weight
646
+ module.decoder.layers.15.self_attention.linear_proj.weight
647
+ module.decoder.final_layernorm.weight
648
+ module.decoder.layers.15.self_attention.linear_qkv.weight
649
+ module.decoder.layers.15.pre_mlp_layernorm.weight
650
+ Params for bucket 3 (41984000 elements, 41984000 padded size):
651
+ module.decoder.layers.14.mlp.router.weight
652
+ module.decoder.layers.14.self_attention.linear_proj.weight
653
+ module.decoder.layers.14.self_attention.linear_qkv.layer_norm_weight
654
+ module.decoder.layers.14.self_attention.linear_qkv.weight
655
+ module.decoder.layers.14.pre_mlp_layernorm.weight
656
+ Params for bucket 4 (41984000 elements, 41984000 padded size):
657
+ module.decoder.layers.13.self_attention.linear_qkv.weight
658
+ module.decoder.layers.13.pre_mlp_layernorm.weight
659
+ module.decoder.layers.13.mlp.router.weight
660
+ module.decoder.layers.13.self_attention.linear_qkv.layer_norm_weight
661
+ module.decoder.layers.13.self_attention.linear_proj.weight
662
+ Params for bucket 5 (41984000 elements, 41984000 padded size):
663
+ module.decoder.layers.12.mlp.router.weight
664
+ module.decoder.layers.12.self_attention.linear_proj.weight
665
+ module.decoder.layers.12.self_attention.linear_qkv.layer_norm_weight
666
+ module.decoder.layers.12.self_attention.linear_qkv.weight
667
+ module.decoder.layers.12.pre_mlp_layernorm.weight
668
+ Params for bucket 6 (41984000 elements, 41984000 padded size):
669
+ module.decoder.layers.11.mlp.router.weight
670
+ module.decoder.layers.11.self_attention.linear_proj.weight
671
+ module.decoder.layers.11.self_attention.linear_qkv.layer_norm_weight
672
+ module.decoder.layers.11.self_attention.linear_qkv.weight
673
+ module.decoder.layers.11.pre_mlp_layernorm.weight
674
+ Params for bucket 7 (41984000 elements, 41984000 padded size):
675
+ module.decoder.layers.10.mlp.router.weight
676
+ module.decoder.layers.10.self_attention.linear_proj.weight
677
+ module.decoder.layers.10.self_attention.linear_qkv.layer_norm_weight
678
+ module.decoder.layers.10.self_attention.linear_qkv.weight
679
+ module.decoder.layers.10.pre_mlp_layernorm.weight
680
+ Params for bucket 8 (41984000 elements, 41984000 padded size):
681
+ module.decoder.layers.9.self_attention.linear_qkv.weight
682
+ module.decoder.layers.9.pre_mlp_layernorm.weight
683
+ module.decoder.layers.9.mlp.router.weight
684
+ module.decoder.layers.9.self_attention.linear_proj.weight
685
+ module.decoder.layers.9.self_attention.linear_qkv.layer_norm_weight
686
+ Params for bucket 9 (41984000 elements, 41984000 padded size):
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.8.mlp.router.weight
690
+ module.decoder.layers.8.self_attention.linear_qkv.weight
691
+ module.decoder.layers.8.pre_mlp_layernorm.weight
692
+ Params for bucket 10 (41984000 elements, 41984000 padded size):
693
+ module.decoder.layers.7.pre_mlp_layernorm.weight
694
+ module.decoder.layers.7.self_attention.linear_qkv.weight
695
+ module.decoder.layers.7.mlp.router.weight
696
+ module.decoder.layers.7.self_attention.linear_proj.weight
697
+ module.decoder.layers.7.self_attention.linear_qkv.layer_norm_weight
698
+ Params for bucket 11 (41984000 elements, 41984000 padded size):
699
+ module.decoder.layers.6.mlp.router.weight
700
+ module.decoder.layers.6.self_attention.linear_qkv.layer_norm_weight
701
+ module.decoder.layers.6.self_attention.linear_qkv.weight
702
+ module.decoder.layers.6.self_attention.linear_proj.weight
703
+ module.decoder.layers.6.pre_mlp_layernorm.weight
704
+ Params for bucket 12 (41984000 elements, 41984000 padded size):
705
+ module.decoder.layers.5.pre_mlp_layernorm.weight
706
+ module.decoder.layers.5.self_attention.linear_proj.weight
707
+ module.decoder.layers.5.self_attention.linear_qkv.weight
708
+ module.decoder.layers.5.mlp.router.weight
709
+ module.decoder.layers.5.self_attention.linear_qkv.layer_norm_weight
710
+ Params for bucket 13 (41984000 elements, 41984000 padded size):
711
+ module.decoder.layers.4.pre_mlp_layernorm.weight
712
+ module.decoder.layers.4.self_attention.linear_qkv.weight
713
+ module.decoder.layers.4.mlp.router.weight
714
+ module.decoder.layers.4.self_attention.linear_proj.weight
715
+ module.decoder.layers.4.self_attention.linear_qkv.layer_norm_weight
716
+ Params for bucket 14 (41984000 elements, 41984000 padded size):
717
+ module.decoder.layers.3.mlp.router.weight
718
+ module.decoder.layers.3.self_attention.linear_qkv.layer_norm_weight
719
+ module.decoder.layers.3.self_attention.linear_proj.weight
720
+ module.decoder.layers.3.self_attention.linear_qkv.weight
721
+ module.decoder.layers.3.pre_mlp_layernorm.weight
722
+ Params for bucket 15 (41984000 elements, 41984000 padded size):
723
+ module.decoder.layers.2.mlp.router.weight
724
+ module.decoder.layers.2.pre_mlp_layernorm.weight
725
+ module.decoder.layers.2.self_attention.linear_qkv.layer_norm_weight
726
+ module.decoder.layers.2.self_attention.linear_qkv.weight
727
+ module.decoder.layers.2.self_attention.linear_proj.weight
728
+ Params for bucket 16 (41984000 elements, 41984000 padded size):
729
+ module.decoder.layers.1.mlp.router.weight
730
+ module.decoder.layers.1.self_attention.linear_qkv.layer_norm_weight
731
+ module.decoder.layers.1.self_attention.linear_proj.weight
732
+ module.decoder.layers.1.self_attention.linear_qkv.weight
733
+ module.decoder.layers.1.pre_mlp_layernorm.weight
734
+ Params for bucket 17 (41984000 elements, 41984000 padded size):
735
+ module.decoder.layers.0.mlp.router.weight
736
+ module.decoder.layers.0.self_attention.linear_proj.weight
737
+ module.decoder.layers.0.self_attention.linear_qkv.weight
738
+ module.decoder.layers.0.pre_mlp_layernorm.weight
739
+ module.decoder.layers.0.self_attention.linear_qkv.layer_norm_weight
740
+ Params for bucket 18 (206045184 elements, 206045184 padded size):
741
+ module.embedding.word_embeddings.weight
742
+ INFO:megatron.core.distributed.param_and_grad_buffer:Number of buckets for gradient all-reduce / reduce-scatter: 7
743
+ Params for bucket 1 (44040192 elements, 44040192 padded size):
744
+ module.decoder.layers.15.mlp.experts.linear_fc2.weight0
745
+ module.decoder.layers.14.mlp.experts.linear_fc1.weight0
746
+ module.decoder.layers.15.mlp.experts.linear_fc1.weight0
747
+ module.decoder.layers.13.mlp.experts.linear_fc2.weight0
748
+ module.decoder.layers.14.mlp.experts.linear_fc2.weight0
749
+ Params for bucket 2 (50331648 elements, 50331648 padded size):
750
+ module.decoder.layers.13.mlp.experts.linear_fc1.weight0
751
+ module.decoder.layers.11.mlp.experts.linear_fc2.weight0
752
+ module.decoder.layers.12.mlp.experts.linear_fc2.weight0
753
+ module.decoder.layers.12.mlp.experts.linear_fc1.weight0
754
+ module.decoder.layers.11.mlp.experts.linear_fc1.weight0
755
+ Params for bucket 3 (44040192 elements, 44040192 padded size):
756
+ module.decoder.layers.10.mlp.experts.linear_fc1.weight0
757
+ module.decoder.layers.9.mlp.experts.linear_fc1.weight0
758
+ module.decoder.layers.10.mlp.experts.linear_fc2.weight0
759
+ module.decoder.layers.8.mlp.experts.linear_fc2.weight0
760
+ module.decoder.layers.9.mlp.experts.linear_fc2.weight0
761
+ Params for bucket 4 (50331648 elements, 50331648 padded size):
762
+ module.decoder.layers.8.mlp.experts.linear_fc1.weight0
763
+ module.decoder.layers.7.mlp.experts.linear_fc1.weight0
764
+ module.decoder.layers.6.mlp.experts.linear_fc2.weight0
765
+ module.decoder.layers.7.mlp.experts.linear_fc2.weight0
766
+ module.decoder.layers.6.mlp.experts.linear_fc1.weight0
767
+ Params for bucket 5 (44040192 elements, 44040192 padded size):
768
+ module.decoder.layers.5.mlp.experts.linear_fc2.weight0
769
+ module.decoder.layers.3.mlp.experts.linear_fc2.weight0
770
+ module.decoder.layers.4.mlp.experts.linear_fc2.weight0
771
+ module.decoder.layers.4.mlp.experts.linear_fc1.weight0
772
+ module.decoder.layers.5.mlp.experts.linear_fc1.weight0
773
+ Params for bucket 6 (50331648 elements, 50331648 padded size):
774
+ module.decoder.layers.1.mlp.experts.linear_fc2.weight0
775
+ module.decoder.layers.2.mlp.experts.linear_fc1.weight0
776
+ module.decoder.layers.3.mlp.experts.linear_fc1.weight0
777
+ module.decoder.layers.2.mlp.experts.linear_fc2.weight0
778
+ module.decoder.layers.1.mlp.experts.linear_fc1.weight0
779
+ Params for bucket 7 (18874368 elements, 18874368 padded size):
780
+ module.decoder.layers.0.mlp.experts.linear_fc2.weight0
781
+ module.decoder.layers.0.mlp.experts.linear_fc1.weight0
782
+ 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 0x7ff22ee005f0>, config_logger_dir='')
783
+ INFO:megatron.core.optimizer_param_scheduler:> learning rate decay style: cosine
784
+ [after model, optimizer, and learning rate scheduler are built] datetime: 2026-08-30 06:43:21
785
+ > building train, validation, and test datasets ...
786
+ > datasets target sizes (minimum size):
787
+ train: 192
788
+ validation: 0
789
+ test: 0
790
+ INFO:megatron.core.datasets.blended_megatron_dataset_config:Let split_matrix = [(0, 1.0), None, None]
791
+ > building train, validation, and test datasets for GPT ...
792
+ INFO:megatron.core.datasets.blended_megatron_dataset_builder:Building GPTDataset splits with sizes=(192, 0, 0) and config=GPTDatasetConfig(random_seed=1234, sequence_length=8192, 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 0x7ff231246e40>, 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)
793
+ INFO:megatron.core.datasets.indexed_dataset:Load the _IndexReader from /workspace/data/wikitext/wikitext103_gpt2_text_document.idx
794
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence lengths
795
+ INFO:megatron.core.datasets.indexed_dataset: Extract the sequence pointers
796
+ INFO:megatron.core.datasets.indexed_dataset: Extract the document indices
797
+ INFO:megatron.core.datasets.indexed_dataset:> total number of sequences: 1165029
798
+ INFO:megatron.core.datasets.indexed_dataset:> total number of documents: 1801350
799
+ INFO:megatron.core.datasets.gpt_dataset:Load the GPTDataset train indices
800
+ INFO:megatron.core.datasets.gpt_dataset: Load the document index from b16a908ae7f33ed67d5ec87f6f633afe-GPTDataset-train-document_index.npy
801
+ INFO:megatron.core.datasets.gpt_dataset: Load the sample index from b16a908ae7f33ed67d5ec87f6f633afe-GPTDataset-train-sample_index.npy
802
+ INFO:megatron.core.datasets.gpt_dataset: Load the shuffle index from b16a908ae7f33ed67d5ec87f6f633afe-GPTDataset-train-shuffle_index.npy
803
+ INFO:megatron.core.datasets.gpt_dataset:> total number of samples: 14536
804
+ > finished creating GPT datasets ...
805
+ [after dataloaders are built] datetime: 2026-08-30 06:43:21
806
+ done with setup ...
807
+ (min, max) time across ranks (ms):
808
+ model-and-optimizer-setup ......................: (237.79, 238.41)
809
+ train/valid/test-data-iterators-setup ..........: (64.57, 70.72)
810
+ training ...
811
+ Overwriting rerun_state_machine.current_iteration from -1 to 0...
812
+ [before the start of training step] datetime: 2026-08-30 06:43:21
813
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
814
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
815
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
816
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
817
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
818
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
819
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
820
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
821
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
822
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
823
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
824
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
825
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
826
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
827
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
828
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
829
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
830
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
831
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
832
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
833
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
834
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
835
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
836
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
837
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
838
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
839
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
840
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
841
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
842
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
843
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
844
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
845
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
846
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
847
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
848
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
849
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
850
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
851
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
852
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
853
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
854
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
855
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
856
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
857
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
858
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
859
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
860
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
861
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
862
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
863
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
864
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
865
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
866
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
867
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
868
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
869
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
870
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
871
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
872
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
873
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
874
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
875
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
876
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
877
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
878
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
879
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
880
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
881
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
882
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
883
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
884
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
885
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
886
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
887
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
888
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
889
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
890
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
891
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
892
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
893
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
894
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
895
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
896
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
897
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
898
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
899
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
900
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
901
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
902
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
903
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
904
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
905
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
906
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
907
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:482: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
908
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
909
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
910
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
911
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
912
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
913
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
914
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
915
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
916
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
917
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
918
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
919
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
920
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
921
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
922
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
923
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:334: UserWarning: The num_dispatched_tokens is deprecated, it will be removed in the future.
924
+ warnings.warn("The num_dispatched_tokens is deprecated, it will be removed in the future.")
925
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
926
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
927
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
928
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
929
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
930
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
931
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
932
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
933
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
934
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
935
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
936
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
937
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
938
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
939
+ /workspace/DeepEP/deep_ep/hybrid_ep_buffer.py:336: UserWarning: The use_host_meta is deprecated, it will be removed in the future.
940
+ warnings.warn("The use_host_meta is deprecated, it will be removed in the future.")
941
+ [2026-08-30 06:43:59] iteration 1/ 12 | consumed samples: 16 | elapsed time per iteration (ms): 38164.0 | throughput per GPU (TFLOP/s/GPU): 5.2 | learning rate: 2.000000E-05 | global batch size: 16 | lm loss: 1.103051E+01 | loss scale: 1.0 | grad norm: 49.674 | number of skipped iterations: 0 | number of nan iterations: 0 |Number of parameters in transformer block in billions: 3.09
942
+
943
+ Number of parameters in embedding layers in billions: 0.41
944
+ Total number of parameters in billions: 3.50
945
+ Number of parameters in most loaded shard in billions: 3.4994
946
+ Theoretical memory footprints: weight and optimizer=25029.43 MB
947
+ [Rank 0] (after 1 iterations) memory (MB) | allocated: 15544.990234375 | max allocated: 31626.068359375 | reserved: 32496.0 | max reserved: 32496.0
948
+ [2026-08-30 06:44:00] iteration 2/ 12 | consumed samples: 32 | elapsed time per iteration (ms): 413.6 | throughput per GPU (TFLOP/s/GPU): 479.6 | learning rate: 4.000000E-05 | global batch size: 16 | lm loss: 1.101006E+01 | loss scale: 1.0 | grad norm: 50.587 | number of skipped iterations: 0 | number of nan iterations: 0 |
949
+ [2026-08-30 06:44:00] iteration 3/ 12 | consumed samples: 48 | elapsed time per iteration (ms): 399.0 | throughput per GPU (TFLOP/s/GPU): 497.2 | learning rate: 6.000000E-05 | global batch size: 16 | lm loss: 1.060427E+01 | loss scale: 1.0 | grad norm: 251.390 | number of skipped iterations: 0 | number of nan iterations: 0 |
950
+ Capture CUDA graph for training!!!
951
+ CUDA graph capture done!!!
952
+ [2026-08-30 06:44:01] iteration 4/ 12 | consumed samples: 64 | elapsed time per iteration (ms): 1288.0 | throughput per GPU (TFLOP/s/GPU): 154.0 | learning rate: 8.000000E-05 | global batch size: 16 | lm loss: 1.310891E+01 | loss scale: 1.0 | grad norm: 40.118 | number of skipped iterations: 0 | number of nan iterations: 0 |
953
+ [2026-08-30 06:44:02] iteration 5/ 12 | consumed samples: 80 | elapsed time per iteration (ms): 317.6 | throughput per GPU (TFLOP/s/GPU): 624.7 | learning rate: 1.000000E-04 | global batch size: 16 | lm loss: 1.053838E+01 | loss scale: 1.0 | grad norm: 19.448 | number of skipped iterations: 0 | number of nan iterations: 0 |
954
+ [2026-08-30 06:44:02] iteration 6/ 12 | consumed samples: 96 | elapsed time per iteration (ms): 320.2 | throughput per GPU (TFLOP/s/GPU): 619.5 | learning rate: 7.750000E-05 | global batch size: 16 | lm loss: 1.033739E+01 | loss scale: 1.0 | grad norm: 8.746 | number of skipped iterations: 0 | number of nan iterations: 0 |
955
+ [2026-08-30 06:44:02] iteration 7/ 12 | consumed samples: 112 | elapsed time per iteration (ms): 316.9 | throughput per GPU (TFLOP/s/GPU): 625.9 | learning rate: 3.250000E-05 | global batch size: 16 | lm loss: 9.515403E+00 | loss scale: 1.0 | grad norm: 7.435 | number of skipped iterations: 0 | number of nan iterations: 0 |
956
+ [2026-08-30 06:44:02] iteration 8/ 12 | consumed samples: 128 | elapsed time per iteration (ms): 316.3 | throughput per GPU (TFLOP/s/GPU): 627.2 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.786578E+00 | loss scale: 1.0 | grad norm: 6.205 | number of skipped iterations: 0 | number of nan iterations: 0 |
957
+ [2026-08-30 06:44:03] iteration 9/ 12 | consumed samples: 144 | elapsed time per iteration (ms): 315.5 | throughput per GPU (TFLOP/s/GPU): 628.7 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.507112E+00 | loss scale: 1.0 | grad norm: 5.330 | number of skipped iterations: 0 | number of nan iterations: 0 |
958
+ [2026-08-30 06:44:03] iteration 10/ 12 | consumed samples: 160 | elapsed time per iteration (ms): 318.0 | throughput per GPU (TFLOP/s/GPU): 623.9 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.418665E+00 | loss scale: 1.0 | grad norm: 4.904 | number of skipped iterations: 0 | number of nan iterations: 0 |
959
+ [2026-08-30 06:44:03] iteration 11/ 12 | consumed samples: 176 | elapsed time per iteration (ms): 317.8 | throughput per GPU (TFLOP/s/GPU): 624.2 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.354745E+00 | loss scale: 1.0 | grad norm: 4.817 | number of skipped iterations: 0 | number of nan iterations: 0 |
960
+ [2026-08-30 06:44:04] iteration 12/ 12 | consumed samples: 192 | elapsed time per iteration (ms): 315.4 | throughput per GPU (TFLOP/s/GPU): 629.0 | learning rate: 1.000000E-05 | global batch size: 16 | lm loss: 8.263580E+00 | loss scale: 1.0 | grad norm: 4.561 | number of skipped iterations: 0 | number of nan iterations: 0 |
961
+ [after training is done] datetime: 2026-08-30 06:44:04
962
+ [rank0]:[W830 06:44:05.681486330 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())
963
+ W0830 06:44:06.094000 648203 torch/distributed/elastic/agent/server/api.py:719] Received 15 death signal, shutting down workers
964
+ W0830 06:44:06.095000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648292 closing signal SIGTERM
965
+ W0830 06:44:06.096000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648293 closing signal SIGTERM
966
+ W0830 06:44:06.097000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648294 closing signal SIGTERM
967
+ W0830 06:44:06.099000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648295 closing signal SIGTERM
968
+ W0830 06:44:06.101000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648296 closing signal SIGTERM
969
+ W0830 06:44:06.102000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648297 closing signal SIGTERM
970
+ W0830 06:44:06.103000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648298 closing signal SIGTERM
971
+ W0830 06:44:06.104000 648203 torch/distributed/elastic/multiprocessing/api.py:900] Sending process 648299 closing signal SIGTERM
972
+ Traceback (most recent call last):
973
+ File "/usr/local/bin/torchrun", line 8, in <module>
974
+ sys.exit(main())
975
+ ^^^^^^
976
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper
977
+ return f(*args, **kwargs)
978
+ ^^^^^^^^^^^^^^^^^^
979
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 892, in main
980
+ run(args)
981
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/run.py", line 883, in run
982
+ elastic_launch(
983
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 139, in __call__
984
+ return launch_agent(self._config, self._entrypoint, list(args))
985
+ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
986
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/launcher/api.py", line 261, in launch_agent
987
+ result = agent.run()
988
+ ^^^^^^^^^^^
989
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/metrics/api.py", line 138, in wrapper
990
+ result = f(*args, **kwargs)
991
+ ^^^^^^^^^^^^^^^^^^
992
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 711, in run
993
+ result = self._invoke_run(role)
994
+ ^^^^^^^^^^^^^^^^^^^^^^
995
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/agent/server/api.py", line 870, in _invoke_run
996
+ time.sleep(monitor_interval)
997
+ File "/usr/local/lib/python3.12/dist-packages/torch/distributed/elastic/multiprocessing/api.py", line 84, in _terminate_process_handler
998
+ raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
999
+ torch.distributed.elastic.multiprocessing.api.SignalException: Process 648203 got signal: 15
tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/results.tsv CHANGED
@@ -2,3 +2,6 @@ config backend policy slots seq gbs cap recompute rep status iter_ms_mean iter_m
2
  PONLY echo one 1 4096 16 2.40625 none 1 OK 175.19 174.10 40157 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq4096__gbs16__cap2p40625/rep1
3
  PONLY echo one 1 4096 16 2.40625 none 2 OK 174.92 173.90 40241 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq4096__gbs16__cap2p40625/rep2
4
  PONLY echo one 1 4096 16 2.40625 none 3 OK 174.81 173.80 40243 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq4096__gbs16__cap2p40625/rep3
 
 
 
 
2
  PONLY echo one 1 4096 16 2.40625 none 1 OK 175.19 174.10 40157 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq4096__gbs16__cap2p40625/rep1
3
  PONLY echo one 1 4096 16 2.40625 none 2 OK 174.92 173.90 40241 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq4096__gbs16__cap2p40625/rep2
4
  PONLY echo one 1 4096 16 2.40625 none 3 OK 174.81 173.80 40243 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq4096__gbs16__cap2p40625/rep3
5
+ PONLY echo one 1 8192 16 2.40625 none 1 OK 316.36 315.50 53279 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep1
6
+ PONLY echo one 1 8192 16 2.40625 none 2 OK 316.47 315.50 53283 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep2
7
+ PONLY echo one 1 8192 16 2.40625 none 3 OK 316.70 315.40 53283 /workspace/tiera-rerun-fixed-moonep/echo/ponly/ponly_gbs16/PONLY__echo__one__seq8192__gbs16__cap2p40625/rep3