DeepSpeed
DeepSpeed Communication Profiling and Logging
#2012
Merged

Commits
  • Staging comms v1 (#301)
    Quentin-Anthony committed 3 years ago
  • Delete stage1.py
    awan-10 committed 3 years ago
  • Delete distributed.py
    awan-10 committed 3 years ago
  • revert deepspeed/__init__.py logging calls
    Quentin-Anthony committed 3 years ago
  • Delete test.py
    Quentin-Anthony committed 3 years ago
  • Update comments and move custom comm ops to internal functions
    Quentin-Anthony committed 3 years ago
  • Merge branch 'staging-comms-next' of https://github.com/microsoft/DeepSpeed into staging-comms-next
    Quentin-Anthony committed 3 years ago
  • Remove unnecessary print and update backend description
    Quentin-Anthony committed 3 years ago
  • Relax assertion to allow Megatron-DeepSpeed MoE to use ZeRO 1
    Quentin-Anthony committed 3 years ago
  • Simplify ZeRO stage 1 check for previous commit
    Quentin-Anthony committed 3 years ago
  • Remove misleading world_size prints
    Quentin-Anthony committed 3 years ago
  • Add commslogger class, and introduce rough prototype comms logging
    Quentin-Anthony committed 3 years ago
  • Clean up logger
    Quentin-Anthony committed 3 years ago
  • Add more robust arg checks
    Quentin-Anthony committed 3 years ago
  • Add labels to common collective calls for logger
    Quentin-Anthony committed 3 years ago
  • Add more annotations
    Quentin-Anthony committed 3 years ago
  • Fix up log_summary_new and fix logging bug for barrier
    Quentin-Anthony committed 3 years ago
  • Clean up arg sweep logic and add isend/irecv
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' into staging-comms-logging-v1
    Quentin-Anthony committed 3 years ago
  • Clean up logging branch
    Quentin-Anthony committed 3 years ago
  • Unify naming and fix circular import
    Quentin-Anthony committed 3 years ago
  • Fix deepspeed comm imports for logging.py
    Quentin-Anthony committed 3 years ago
  • Added comms config support, removed some log names
    Quentin-Anthony committed 3 years ago
  • Add comms config file
    Quentin-Anthony committed 3 years ago
  • Add pydantic to requirements
    Quentin-Anthony committed 3 years ago
  • Add configure non-op to old torch
    Quentin-Anthony committed 3 years ago
  • Update logging call for old torch
    Quentin-Anthony committed 3 years ago
  • Add log_name placeholder args for old torch
    Quentin-Anthony committed 3 years ago
  • Add basic verbosity setup
    Quentin-Anthony committed 3 years ago
  • Complete verbosity setup
    Quentin-Anthony committed 3 years ago
  • move comms logging to separate file and clean up
    Quentin-Anthony committed 3 years ago
  • Change debug message design
    Quentin-Anthony committed 3 years ago
  • refactor debug helper and clean up
    Quentin-Anthony committed 3 years ago
  • Refactor a bit and clean up prints
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' into staging-comms-logging-v1
    Quentin-Anthony committed 3 years ago
  • config docs, remove old log_summary func, fix imports
    Quentin-Anthony committed 3 years ago
  • Finished docs, added import, fixed non-debug calls
    Quentin-Anthony committed 3 years ago
  • Ran pre-commit
    Quentin-Anthony committed 3 years ago
  • Removed old comments
    Quentin-Anthony committed 3 years ago
  • Updated fn signatures for torch1.2
    Quentin-Anthony committed 3 years ago
  • Remove lingering prof arg
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' into staging-comms-logging-v1
    jeffra committed 3 years ago
  • Update logging tutorial
    Quentin-Anthony committed 3 years ago
  • Removed unnecessary imports and cleaned up comments
    Quentin-Anthony committed 3 years ago
  • Take master's cleaner comms init logic
    Quentin-Anthony committed 3 years ago
  • Fixed bw calculations and made all logging calls blocking
    Quentin-Anthony committed 3 years ago
  • Added comms logging synch disclaimer
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' into staging-comms-logging-v1
    Quentin-Anthony committed 3 years ago
  • Added using_mpi flag for logging
    Quentin-Anthony committed 3 years ago
  • Formatting
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' of https://github.com/microsoft/DeepSpeed into staging-comms-logging-v1
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' into staging-comms-logging-v1
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' into staging-comms-logging-v1
    Quentin-Anthony committed 3 years ago
  • Merge branch 'master' into staging-comms-logging-v1
    Quentin-Anthony committed 3 years ago
Loading