85 Commits (ce3c4eca7bc2c5b148dfe5db1ddb702558af4831)

Author	SHA1	Message	Date
Jiarui Fang	b87496a66b	[hotfix] fix auto policy of test_sharded_optim_v2 (#2157)	2 years ago
Jiarui Fang	ee287620f0	[Gemini] revert ZeROInitCtx related tracer (#2138)	2 years ago
Jiarui Fang	c89c66a858	[Gemini] update API of the chunkmemstatscollector. (#2129)	2 years ago
Jiarui Fang	2938edf446	[Gemini] update the non model data record method in runtime memory tracer (#2128)	2 years ago
Jiarui Fang	8fac837679	[Gemini] update non model data calculation method (#2126)	2 years ago
Jiarui Fang	5efda69735	[Gemini] hotfix the unittest bugs (#2125)	2 years ago
Jiarui Fang	05bb28aacf	[Gemini] mapping of preop timestep and param (#2124)	2 years ago
Jiarui Fang	9214d1fe28	[Gemini] chunk init using runtime visited param order (#2115)	2 years ago
Jiarui Fang	e5aa8333e4	[NFC] update chunk manager API (#2119)	2 years ago
Jiarui Fang	e99edfcb51	[NFC] polish comments for Chunk class (#2116)	2 years ago
Jiarui Fang	8afc001f4f	[Gemini] chunk init use OrderedParamGenerator (#2110)	2 years ago
HELSON	63fbba3c19	[zero] add L2 gradient clipping for ZeRO (#2112) ... * [zero] add L2 gradient clipping * [testing] add MlpModel * [zero] add unit test for grad clipping * fix atol	2 years ago
Jiarui Fang	70a8556946	[gemini] get the param visited order during runtime (#2108)	2 years ago
Jiarui Fang	61f31c3cf0	[Gemini] NFC, polish search_chunk_configuration (#2107)	2 years ago
Jiarui Fang	85efb7ac2e	[Gemini] gemini use the runtime memory tracer (RMT) (#2099)	2 years ago
Jiarui Fang	4b055351b0	[Gemini] make RuntimeMemTracer work correctly (#2096)	2 years ago
Jiarui Fang	1fca5d79ea	[Gemini] remove GLOBAL_MODEL_DATA_TRACER (#2091)	2 years ago
Jiarui Fang	28e55c2530	[Gemini] remove GLOBAL_CUDA_MEM_INFO (#2090)	2 years ago
Jiarui Fang	25abae6d7f	[Gemini] use MemStats in Runtime Memory tracer (#2088)	2 years ago
Jiarui Fang	33f4412102	[Gemini] use MemStats to store the tracing data. Seperate it from Collector. (#2084)	2 years ago
Jiarui Fang	1f99205827	[Gemini] remove static tracer (#2083)	2 years ago
Jiarui Fang	b3b89865e2	[Gemini] ParamOpHook -> ColoParamOpHook (#2080)	2 years ago
Jiarui Fang	a7adad9ccb	[Gemini] rename hooks related to runtime mem tracer (#2076)	2 years ago
Jiarui Fang	223332ff7e	[Gemini] rename ParamTracerWrapper -> RuntimeMemTracer (#2073)	2 years ago
Jiarui Fang	9f828ef36f	[Gemini] remove not used MemtracerWrapper (#2072)	2 years ago
Zihao	38ea4ba1bd	[Gemini] fix grad unreleased issue and param recovery issue (#2052)	2 years ago
Zihao	6a9158f1fa	[Gemini] free and allocate cuda memory by tensor.storage, add grad hook (#2040)	2 years ago
Jiarui Fang	28aa9a4294	[Gemini] more rigorous unit tests for run_fwd_bwd (#2034)	2 years ago
Zihao	95c4532fff	[Gemini] paramWrapper paramTracerHook unitest (#2030)	2 years ago
Jiarui Fang	8daf1b4db1	[Gemini] patch for supporting orch.add_ function for ColoTensor (#2003)	2 years ago
Zihao	a719b89a41	[gemini] param_trace_hook (#2020)	2 years ago
Jiarui Fang	0b0d8f9e17	[hotfix] revert bug PRs (#2016)	2 years ago
Zihao	aba3db464d	[Gemini] ParamMemHook (#2008)	2 years ago
Zihao	0160a62a3c	[Gemini] param_tracer_wrapper and test case (#2009)	2 years ago
Jiarui Fang	3712ac7f90	[Gemini] add bert for MemtracerWrapper unintests (#1982)	2 years ago
Jiarui Fang	e481489aa6	[Gemini] MemtracerWrapper unittests (#1981)	2 years ago
Jiarui Fang	31922110ad	[Gemini] memory trace hook (#1978)	2 years ago
Jiarui Fang	0529fcde06	[Gemini] independent runtime tracer (#1974)	2 years ago
Jiarui Fang	7e24b9b9ee	[Gemini] clean no used MemTraceOp (#1970)	2 years ago
Jiarui Fang	c4739a725a	[Gemini] polish memstats collector (#1962)	2 years ago
Zihao	20e255d4e8	MemStatsCollectorStatic (#1765)	2 years ago
Jiarui Fang	c248800359	[kernel] skip tests of flash_attn and triton when they are not available (#1798)	2 years ago
HELSON	c6a1a62636	[hotfix] fix zero's incompatibility with checkpoint in torch-1.12 (#1786) ... * [hotfix] fix zero's incompatibility with checkpoint in torch-1.12 * [zero] add cpu shard init * [zero] add tiny example test * [colo_tensor] fix bugs for torch-1.11	2 years ago
Jiarui Fang	cb5a587e9a	[hotfix] polish chunk import (#1787)	2 years ago
Jiarui Fang	f34dab4270	[compatibility] ChunkMgr import error (#1772)	2 years ago
HELSON	f69f9bf223	[zero] add chunk init function for users (#1729) ... * add chunk manager init function * fix unit tests * add comment * add flush=True	2 years ago
HELSON	1468e4bcfc	[zero] add constant placement policy (#1705) ... * fixes memory leak when paramter is in fp16 in ZeroDDP init. * bans chunk releasement in CUDA. Only when a chunk is about to offload, it is allowed to release. * adds a constant placement policy. With it, users can allocate a reserved caching memory space for parameters.	2 years ago
HELSON	b28991dd0a	[feature] A new ZeRO implementation (#1644)	2 years ago
Jiarui Fang	c5d39215f6	Revert "[feature] new zero implementation (#1623)" (#1643) ... This reverts commit 5be118f405.	2 years ago
HELSON	5be118f405	[feature] new zero implementation (#1623)	2 years ago