Paddle

Commit Graph

Author	SHA1	Message	Date
chengduo	21440b4d69	Add call stack info during compile time (#19067 ) * Add call stack info during runtime and compile time test=develop * Rename operator_call_stack test=develop * Add unit test test=develop * follow comment test=develop	6 years ago
jiaqi	fc038da749	fix QueueDataset queue size (#19016 ) * fix QueueDataset queue size，set queue size = batch size * 100, to avoid too many instances in channel when training is much slower than reading data.	6 years ago
Leo Chen	8f53735437	Fix memory overwriting of tensors returned by executor (#19030 ) * fix memory overlapping of fetch var (return of executor.run), test=develop * fix wrong usage of ParallelExecutor in op_test, test=develop * remove useless parameter and simplify code * avoid tensor destruct untimely, test=develop * add testcase independent of OpTest, test=develop	6 years ago
Zeng Jinle	2175d19993	fix memory_reuse_pass memory_size calculation error, test=develop (#19020 )	6 years ago
Zeng Jinle	7ac748adb4	Open gc by default (#18836 ) * open gc by default, test=develop * fix test_train_recognize_digits and disable gc when ngraph is enabled, test=develop * fix conditional_block op eager deletion bug, test=develop * add some comments to reviewers, test=develop	6 years ago
jiaqi	02c370c3dc	support filelist size < trainer num && fix pull dense (#18956 ) * support filelist size < trainer num * pull dense when stop, to make sure local dense params are same as pserver, so save paddle model will save dense model same as pserver * enable QueueDataset train same filelist for serveral times	6 years ago
chengduo	e7da0940f9	Disable fuse optimization option (#18924 ) * Disable fuse optimization test=develop	6 years ago
石晓伟	ee2f296ef8	Fusion: seqpool_cvm_concat (#18471 ) * add fusion_seqpool_cvm_concat test=develop * simplify pass, test=develop * fix code style, test=develop	6 years ago
jiaqi	768059b3a0	adjust ins weight according to nid slot (#18784 ) adjust ins weight according to nid slot , user can specify adjust_ins_weight in strategy	6 years ago
Leo Zhao	10eeed93d1	Revert "use static variable to do cache instead of thread local in thread frequent switching case (#18428 )" (#18879 ) This reverts commit `ce38bb5341`. test=develop	6 years ago
Zeng Jinle	8008ab4e6b	Remove legacy C++ memory optimization codes (#18834 ) * remove legacy memory optimization codes, test=develop * follow huihuang's comments,test=develop * follow luotao's comments, test=develop	6 years ago
Thunderbrook	52c1431eee	add clear_model interface in fleetwrapper (#18815 ) * dump slot * test * proto * dump slot * test * proto * code style * code style * code style * style * add delete after unseen days * add unseen days * code style * conflict solve test=develop * add clear model * code style test=develop * code style test=develop	6 years ago
chengduo	4140fe11a4	Open fuse optimization ops (#18741 ) * open fuse optimization ops test=develop	6 years ago
Zeng Jinle	a802da650b	Feature/mem opt pass refactor (#18735 ) * first version memory optimize pass, test=develop * remove move_tensor_sharing_pass, test=develop * refine code comments, add unittests, test=develop * turn off memory_optimize by default, test=develop * follow huihuang's comments, test=develop * follow chengduoZH's comments, test=develop * fix grammar error, add const qualifier, fix pass_test exception message, test=develop * follow chengduoZH's comments 2nd, test=develop	6 years ago
fuyinno4	c167a4b4dd	Fix shrink-dense and add scale-datanorm (#18746 ) Fix FleetWrapper: 1. fix shrink dense: just scale show 2. add datanorm scale: divide datanorm's gradient by batch_size	6 years ago
Zhaolong Xing	26ae6d49e4	Update trt5 for paddle-trt (#18645 ) * update paddle-trt for: 1. fix bug: when batch > 2, core in split plugin. 2. add leaky_relu trt5.0 support (yolov3 from 65ms to 42ms.) 3. add new attr to dropout. 4. shuffle channel, swish, relu6 support test=develop * 1. fix ci test=develop	6 years ago
Thunderbrook	d8396281ef	add slot to sparse table (#18686 ) The change includes 2 things: 1. save delta model and shrink table are control by the same parameter before, now add delete_after_unseen_days to control shrink table. 2. value in sparse table has no slot before, now add slot in sparse table, and add DownpureCtrAccessor to support the new meta. test=develop	6 years ago
jiaqi	d18aabb472	support patch data, add load_one_table, fix bug (#18509 ) （1）support patch data （merge slots of instances of same line id, modify dense layer which changes its size）（2）add fleet load_one_table interface, support load from paddle model and load from pslib model （3）fix push sparse bug which cause push sparse cost more time（about 10% in my testcase）（4）when some slots are not in one of your network (join/update, etc.)，data feed、collect label info、push/pull sparse will skip these slots， instead of throw error. （5）add more debug info in TrainFilesWithProfiler	6 years ago
chengduo	fd3aad6cb3	Make fuse_optimizer_op_pass also work when the model contains sparse gradients. (#18664 ) * support sparse gradients test=develop	6 years ago
Huihuang Zheng	89bc3fd841	Support memory eager deletion on recurrent OP (#17710 ) Test PaddingRNN on V100 GPU device. Test configuration: large model, padding mode (which is the mode using recurrentOp), one GPU. GPU memory (MiB): 6414 (this PR) vs 6837 (without this PR) Speed (steps/s): 10.28 (this PR) vs 9.89 (without this PR)	6 years ago
Zeng Jinle	ae58afc546	Feature/auto_growth_allocator (#18561 ) * feature/auto_growth_allocator, test=develop * add unittest of AlignedAllocator, test=develop * try to turn on auto_growth to test on CI, test=develop * fix segmentation fault in mixed_vector.h, test=develop * add unittests, test=develop	6 years ago
guru4elephant	d714bf037c	remove async executor and add data_feed.proto to the deps of train demo (#18659 ) * remove async executor and add data_feed.proto to the deps of train demo	6 years ago
chengduo	a6d468a265	fix PE fetch bug (#18644 ) test=develop	6 years ago
Leo Zhao	ff77dea969	not use transferscope cache in cpu case (#18578 ) * not use transferscope cache in cpu case test=develop * adjust variable name and add comments test=develop * use correct format for class member in operator.h * use correct format for class member in operator.cc test=develop	6 years ago
123malin	b414645a65	fix #17430 : int64类型的attr训练非预期 (#18264 ) * fix int64_t * update fill constant op unittest * add empty line	6 years ago
gongweibao	c0a82748cf	Polish backwards optimizer dependency codes and use more default values. (#18255 )	6 years ago
Zeng Jinle	d3003a1620	Feature/buffer_shared_inplace (#17911 ) * feature/buffer_shared_inplace, test=develop * refine code, test=develop * fix elementwise_add op cpu inplace and sum inplace bug, test=develop * add unittest and debug log, test=develop * fix parallel_executor scope bug, polish code, test=develop * fix sum op, activation op, single_in_place_inference bug, test=develop * remove kLocalExecScopeName, test=develop * fix unittest,test=develop * fix out_var first version bug, test=develop * follow comments,test=develop	6 years ago
Zeng Jinle	be24e5b391	Clean unused code of dim and place (#18565 ) * clean code of dim and place, test=develop * fix failed unittests, test=develop	6 years ago
Jiabin Yang	667f88f9a6	Fix/gcc 4.8 ubt link error (#18558 ) * test=develop, fix docker with paddle nccl problem * test=develop, fix/gcc_4.8_ubt_link_error * test=develop, fix code format	6 years ago
Zhaolong Xing	88b52a27fe	Inference: fix mask rcnn model diff, optim memory usage, memory leak. (#18532 ) * Fix Mask rcnn predictor 1. refine memory optim algorithm to support the model with the block op. 2. output diff : modify the affine channel fuse 3. add condition_block_infer op add interface for setting trt calib table dir test=develop * add the missing files. test=develop	6 years ago
Leo Zhao	ce38bb5341	use static variable to do cache instead of thread local in thread frequent switching case (#18428 )	6 years ago
gongweibao	160ddc980c	Regroup fusion by date type. (#18496 )	6 years ago
chengduo	7453857324	Make fuse_all_reduce_op_pass support mix_precision (#17652 )	6 years ago
pkpk	e9c7e218f2	Nan debugger init (#18401 ) test=develop	6 years ago
Tao Luo	d234aa02cd	add transfer_scope_cache unit-test (#18467 ) test=develop	6 years ago
Yi Liu	a873fa84ce	supports collective training with programs (#18392 ) 1. Since allreduce op has 4 reduce types, We split these four reduce types into four ops 2. We also refined the collective op code, e.g. we separated the collective op kernel into CPUKernel and CUDAKernel, and remove the device specified DeviceContext parameter in template as we already knew the target DeviceContext 3. We remove the newly added Collective op role to reduce the complexity of program and graph analysis	6 years ago
Michał Gallus	7023a86c3a	Fix Pooling output scale (#18186 ) * Int8: Fix Pooling output scale test=develop * Update scales quantization for certain operators These include: concat, transpose, pool and reshape. test=develop * Move concat minimum scale finding to quantizer test=develop	6 years ago
jiaqi	93a2b317f7	fix data feed ptr error (#18419 ) fix data feed ptr runtime error, pipeline trainer will core in some cases, so set it nullptr as default value.	6 years ago
chengduo	8ed33bf91f	Fix Bug-prone code of PE (#18354 ) * update pe reduce config test=develop * drop the local_exe_scopes of the previous parallel_executor test=develop	6 years ago
tangwei12	999d9a59a5	fix communicator with pyreader (#18350 ) * add is_runnning in communicator, test=develop	6 years ago
HaoRen	b7128bac5f	supports collective communicated training (#18175 ) * fix prepare context redundant code problem, optimize executor by caching create_varaiables test=develop * supports collective training in executor * make fetch_list runable with variables, add more unittest for use_program_cache test=develop * fix comment test=develop * use unique name for nccl_id * supports output to stream in program_to_code * insert sync_comm_stream before regularization; add skip_op_callstack capability in program_to_code * set op role in collective training * add collective op role * remove orig file * add build optimizer by strategy * add collective strategy * refine collective strategy * add multi-process role maker * refine strategy building factory so that we can easily plugin more strategy * scale loss grad in collective sgd transpiler * add support for distributed fc * code format * revert some features for dist fc * add support for distributed fc training * fix prepare context redundant code problem, optimize executor by caching create_varaiables test=develop * supports collective training in executor * make fetch_list runable with variables, add more unittest for use_program_cache test=develop * use unique name for nccl_id * supports output to stream in program_to_code * insert sync_comm_stream before regularization; add skip_op_callstack capability in program_to_code * set op role in collective training * add collective op role * fix comment test=develop * remove orig file * add build optimizer by strategy * add collective strategy * refine collective strategy * add multi-process role maker * refine strategy building factory so that we can easily plugin more strategy * scale loss grad in collective sgd transpiler * add support for distributed fc * code format * revert some features for dist fc * add support for distributed fc training * test=develop add collective op unittest standard * test=develop remove the test_collective directory * test=develop remove the test_collective directory * remove slicegather test * code format for reducescatter * update attr of shard_index_op * Modify macro nccl_helper * remove test without distribute * macro collective_helper * marcro update * test=develop update support python3.5 * test=develop change gpu memory use to 0.1 when test * test=develop update ut equal func * test=develop set flags to 1.5 * test=develop fix pickle dumple py35 * test=develop fix divide in slice and add sync_comm_stream update atol and rtol to 1e-05 rm shard_index op and test modify read input from file to read from memory remove origin_program in framework and add i/o in c_sync_calc_stream * test=develop update unittest sync operator I/O	6 years ago
Sylwester Fraczek	9252e8fa08	add int8 mkldnn prior_box (#17242 ) add prior_box quantization code add scale algo rules for prior box test=develop	6 years ago
chengduo	135a59ed45	update reduce config (#18334 ) test=develop	6 years ago
chengduo	5489216eba	Clean build strategy (#18148 ) * clean build_strategy test=develop * DataBalanceOpHandle has been removed test=develop * debug * update build_strategy. test=develop	6 years ago
chengduo	14e1e165df	update alloc_continuous_space_for_grad_pass (#18287 ) test=develop	6 years ago
jiaqi	3f8031e256	dataset (#17973 ) (1) use channel instead of vector/BlockingQueue in Dataset，to keep same with existing implementation, and make code more readable and flexible (dataset single output channel or multi output channel). one previous memory out of limit problem is cause by not release memory after training. (2) add Record because MultiSlotType costs too much memory (80B)，fix memory out of limit problem. (3) add Channel, Archive in paddle/fluid/framework (4) change dataset from shared_ptr to unique_ptr in pybind (5) move create/destroy readers from trainer to dataset (6) move shuffle from datafeed to dataset. dataset holds memory, datafeed is only for load data and feed data to network. (7) fix thread num bug of Dataset when filelist size < thread num (8) support set_queue_num in InMemoryDataset	6 years ago
chengduo	25f3cd6486	Update execution_strategy option default value (#18183 ) * update execution_strategy option default value test=develop * fix doc error test=develop	6 years ago
chengduo	4978db2c10	Remove nccl dep when the number of GPU is 1 (#18158 ) * remove nccl dep when the number of GPU is 1 test=develop	6 years ago
chengduo	24e988a471	Fix bug of scope_buffered_ssa_graph_executor (#18100 ) * fix code bug test=develop	6 years ago
gongweibao	f5caf3443c	Fix reinitialized ncclid error! (#18025 )	6 years ago

1 2 3 4 5 ...

2511 Commits (21440b4d69d1a8b390ae62611b2ea0e90261e14c)