File size: 29,853 Bytes
f291f93
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
/home/dat/pino/lib/python3.8/site-packages/jax/lib/xla_bridge.py:382: UserWarning: jax.host_count has been renamed to jax.process_count. This alias will eventually be removed; please update your code.
  warnings.warn(
/home/dat/pino/lib/python3.8/site-packages/jax/lib/xla_bridge.py:369: UserWarning: jax.host_id has been renamed to jax.process_index. This alias will eventually be removed; please update your code.
  warnings.warn(
Epoch ... (1/5):   0%|                                                     | 0/5 [00:00<?, ?it/s]
Epoch ... (1/5):   0%|                                                     | 0/5 [02:13<?, ?it/s]
Traceback (most recent call last):
  File "./run_mlm_flax.py", line 725, in <module>
    state, train_metric, dropout_rngs = p_train_step(state, model_inputs, dropout_rngs)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/_src/traceback_util.py", line 183, in reraise_with_filtered_traceback
    return fun(*args, **kwargs)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/_src/api.py", line 1647, in f_pmapped
    out = pxla.xla_pmap(
  File "/home/dat/pino/lib/python3.8/site-packages/jax/core.py", line 1620, in bind
    return call_bind(self, fun, *args, **params)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/core.py", line 1551, in call_bind
    outs = primitive.process(top_trace, fun, tracers, params)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/core.py", line 1623, in process
    return trace.process_map(self, fun, tracers, params)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/core.py", line 606, in process_call
    return primitive.impl(f, *tracers, **params)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/interpreters/pxla.py", line 624, in xla_pmap_impl
    compiled_fun, fingerprint = parallel_callable(fun, backend, axis_name, axis_size,
  File "/home/dat/pino/lib/python3.8/site-packages/jax/linear_util.py", line 262, in memoized_fun
    ans = call(fun, *args)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/interpreters/pxla.py", line 899, in parallel_callable
    compiled = xla.backend_compile(backend, built, compile_options)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/interpreters/xla.py", line 360, in backend_compile
    return backend.compile(built_c, compile_options=options)
jax._src.traceback_util.UnfilteredStackTrace: RuntimeError: Resource exhausted: Ran out of memory in memory space hbm. Used 35.77G of 15.48G hbm. Exceeded hbm capacity by 20.29G.
Total hbm usage >= 36.29G:
    reserved        530.00M
    program          35.77G
    arguments            0B
Output size 0B; shares 0B with arguments.
Program hbm requirement 35.77G:
    global           692.0K
    scoped           253.0K
    HLO temp         35.77G (97.6% utilization: Unpadded (34.82G) Padded (35.67G), 0.3% fragmentation (105.77M))
  Largest program allocations in hbm:
  1. Size: 6.15G
     Operator: op_type="dot_general" op_name="pmap(train_step)/dot_general[ dimension_numbers=(((2,), (0,)), ((), ()))\n                              precision=None\n                              preferred_element_type=None ]" source_file="/home/dat/pino/lib/python3.8/site-packages/flax/linen/linear.py" source_line=175
     Shape: f32[8,4096,50358]{1,2,0:T(8,128)}
     Unpadded size: 6.15G
     Extra memory due to padding: 256.0K (1.0x expansion)
     XLA label: %fusion.1737.remat4 = f32[8,4096,50358]{1,2,0:T(8,128)} fusion(f32[50358]{0:T(1024)} %get-tuple-element.23321, f32[768,50358,1]{0,1,2:T(8,128)} %bitcast.5512, f32[768]{0:T(1024)} %get-tuple-element.23322, f32[768]{0:T(1024)} %get-tuple-element.23323, f32[8...
     Allocation type: HLO temp
     ==========================
  2. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1805.remat6 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2407, f32[8,12,28,128]{3,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  3. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13201 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2412, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9402, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  4. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1805.remat6 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2407, f32[8,12,28,128]{3,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  5. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13202 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2411, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9401, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  6. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1814 = bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)} fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2416, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9406, f32[8,12,28,128,128]{3,4,2,1,0:T(8,128)} %get-tuple-element.20627, f32[8,12,28,128,384]{...
     Allocation type: HLO temp
     ==========================
  7. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13199 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2414, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9404, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  8. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13200 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2413, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9403, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  9. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1816.remat = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2418, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9408, f32[8,12,28,128,128]{3,4,2,1,0:...
     Allocation type: HLO temp
     ==========================
  10. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1815.remat = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2417, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9407, f32[8,12,28,128,128]{3,4,2,1,0:...
     Allocation type: HLO temp
     ==========================
  11. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13203 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2410, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9400, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  12. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13204 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2409, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9399, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  13. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13205 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2408, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9398, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  14. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11557 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25239, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28505.remat_uncompressed, f32[8,12...
     Allocation type: HLO temp
     ==========================
  15. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11549 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25240, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28470.remat_uncompressed.remat, f3...
     Allocation type: HLO temp
     ==========================
  16. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11469 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.20990, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28115, f32[8,12,32,128,64]{3,2,4,1...
     Allocation type: HLO temp
     ==========================
  17. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11477 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.20989, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28151, f32[8,12,32,128,64]{3,2,4,1...
     Allocation type: HLO temp
     ==========================
  18. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11541 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25236, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28435.remat_uncompressed, f32[8,12...
     Allocation type: HLO temp
     ==========================
  19. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=584
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.2085.remat5.1.remat = f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)} fusion(f32[8,28,128,384]{2,3,1,0:T(8,128)} %get-tuple-element.20992, bf16[8,12,28,384,64]{3,2,4,1,0:T(8,128)(2,1)} %fusion.2473.remat_uncompressed, f32[8,12,32,128,64]{3,2,4,1,0:T(8,128...
     Allocation type: HLO temp
     ==========================
  20. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11533 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25238, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28400.remat_uncompressed, f32[8,12...
     Allocation type: HLO temp
     ==========================
The stack trace below excludes JAX-internal frames.
The preceding is the original exception that occurred, unmodified.
--------------------
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
  File "./run_mlm_flax.py", line 725, in <module>
    state, train_metric, dropout_rngs = p_train_step(state, model_inputs, dropout_rngs)
  File "/home/dat/pino/lib/python3.8/site-packages/jax/interpreters/xla.py", line 360, in backend_compile
    return backend.compile(built_c, compile_options=options)
RuntimeError: Resource exhausted: Ran out of memory in memory space hbm. Used 35.77G of 15.48G hbm. Exceeded hbm capacity by 20.29G.
Total hbm usage >= 36.29G:
    reserved        530.00M
    program          35.77G
    arguments            0B
Output size 0B; shares 0B with arguments.
Program hbm requirement 35.77G:
    global           692.0K
    scoped           253.0K
    HLO temp         35.77G (97.6% utilization: Unpadded (34.82G) Padded (35.67G), 0.3% fragmentation (105.77M))
  Largest program allocations in hbm:
  1. Size: 6.15G
     Operator: op_type="dot_general" op_name="pmap(train_step)/dot_general[ dimension_numbers=(((2,), (0,)), ((), ()))\n                              precision=None\n                              preferred_element_type=None ]" source_file="/home/dat/pino/lib/python3.8/site-packages/flax/linen/linear.py" source_line=175
     Shape: f32[8,4096,50358]{1,2,0:T(8,128)}
     Unpadded size: 6.15G
     Extra memory due to padding: 256.0K (1.0x expansion)
     XLA label: %fusion.1737.remat4 = f32[8,4096,50358]{1,2,0:T(8,128)} fusion(f32[50358]{0:T(1024)} %get-tuple-element.23321, f32[768,50358,1]{0,1,2:T(8,128)} %bitcast.5512, f32[768]{0:T(1024)} %get-tuple-element.23322, f32[768]{0:T(1024)} %get-tuple-element.23323, f32[8...
     Allocation type: HLO temp
     ==========================
  2. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1805.remat6 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2407, f32[8,12,28,128]{3,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  3. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13201 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2412, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9402, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  4. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1805.remat6 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2407, f32[8,12,28,128]{3,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  5. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13202 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2411, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9401, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  6. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1814 = bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)} fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2416, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9406, f32[8,12,28,128,128]{3,4,2,1,0:T(8,128)} %get-tuple-element.20627, f32[8,12,28,128,384]{...
     Allocation type: HLO temp
     ==========================
  7. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13199 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2414, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9404, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  8. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13200 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2413, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9403, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  9. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1816.remat = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2418, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9408, f32[8,12,28,128,128]{3,4,2,1,0:...
     Allocation type: HLO temp
     ==========================
  10. Size: 672.00M
     Operator: op_type="div" op_name="pmap(train_step)/div" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=619
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.1815.remat = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2417, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9407, f32[8,12,28,128,128]{3,4,2,1,0:...
     Allocation type: HLO temp
     ==========================
  11. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13203 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2410, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9400, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  12. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13204 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2409, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9399, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  13. Size: 672.00M
     Shape: bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}
     Unpadded size: 672.00M
     XLA label: %fusion.13205 = (bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}, bf16[8,12,28,128,1024]{3,4,2,1,0:T(8,128)(2,1)}) fusion(f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.2408, f32[8,12,28,128]{3,2,1,0:T(8,128)} %fusion.9398, f32[8,12,28,128,128]{3,4,2,1,0:T(8,1...
     Allocation type: HLO temp
     ==========================
  14. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11557 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25239, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28505.remat_uncompressed, f32[8,12...
     Allocation type: HLO temp
     ==========================
  15. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11549 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25240, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28470.remat_uncompressed.remat, f3...
     Allocation type: HLO temp
     ==========================
  16. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11469 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.20990, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28115, f32[8,12,32,128,64]{3,2,4,1...
     Allocation type: HLO temp
     ==========================
  17. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11477 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.20989, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28151, f32[8,12,32,128,64]{3,2,4,1...
     Allocation type: HLO temp
     ==========================
  18. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11541 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25236, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28435.remat_uncompressed, f32[8,12...
     Allocation type: HLO temp
     ==========================
  19. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=584
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.2085.remat5.1.remat = f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)} fusion(f32[8,28,128,384]{2,3,1,0:T(8,128)} %get-tuple-element.20992, bf16[8,12,28,384,64]{3,2,4,1,0:T(8,128)(2,1)} %fusion.2473.remat_uncompressed, f32[8,12,32,128,64]{3,2,4,1,0:T(8,128...
     Allocation type: HLO temp
     ==========================
  20. Size: 504.00M
     Operator: op_type="dot_general" op_name="pmap(train_step)/jit(jvp(_einsum))/dot_general[ dimension_numbers=(((4,), (4,)), ((0, 1, 2), (0, 1, 2)))\n                                                precision=None\n                                                preferred_element_type=None ]" source_file="/home/dat/transformers/src/transformers/models/big_bird/modeling_flax_big_bird.py" source_line=591
     Shape: f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}
     Unpadded size: 504.00M
     XLA label: %fusion.11533 = (f32[8,12,28,128]{3,2,1,0:T(8,128)}, f32[8,12,28,128,384]{3,4,2,1,0:T(8,128)}) fusion(s32[8,12,30,128,384]{3,4,2,1,0:T(8,128)} %get-tuple-element.25238, bf16[8,12,28,384,64]{3,2,1,0,4:T(8,128)(2,1)} %slice.28400.remat_uncompressed, f32[8,12...
     Allocation type: HLO temp
     ==========================