Pull Requests deepspeedai/DeepSpeed

Fix for lm head weights for models such as llama

#4088 opened 2023-08-04 07:50 by puneeshkhanna

llama with HE

#4087 opened 2023-08-04 03:15 by ciayomin

Fix module parameter replacement for HybridEngine

#4077 opened 2023-08-01 19:59 by XuehaiPan

fix bloom bugs about inference and TP

#3973 opened 2023-07-17 14:01 by dawson-chen

Fix the call to get_param_coordinator() in _end_of_forward_hook()

#3972 opened 2023-07-17 12:22 by mmhab

autoTP for HE

#3957 opened 2023-07-13 23:30 by molly-smith

Pipe Engine Reduce High Dimension Output tensor fix

#3774 opened 2023-06-20 19:46 by abhilash1910

Update vae.py

#3761 opened 2023-06-16 15:16 by mzamini92

Adding assertion for mp_group in HE.

#3740 opened 2023-06-12 17:44 by jomayeri

Addressing ipg Buffer Data Race Condition in Zero Stage2

#3727 opened 2023-06-09 11:17 by xxr3376

Asymmetric quant algorithm update

#3696 opened 2023-06-06 21:48 by cmikeh2

Add FALCON-40B Inference-Kernel Support

#3656 opened 2023-06-01 08:11 by RezaYazdaniAminabadi

Fix optimizer saving with BF16 + PP + ZeRO-1

#3555 opened 2023-05-17 04:04 by Sengxian

Resolve hard dependency on MOE for contiguous_gradients on stage 1

#3545 opened 2023-05-15 13:20 by abhilash1910

Extend DeepSpeed inference initialization API with a 'quantize_groups' argument

#3519 opened 2023-05-11 17:51 by sakogan

Bing/optimizer naming

#3354 opened 2023-04-23 02:02 by xiexbing

Fix handling of (CUDA,ROCR)_VISIBLE_DEVICES

#3165 opened 2023-04-08 20:52 by jglaser

ckpt: add open/close for reading checkpoint

#3009 opened 2023-03-13 22:00 by adammoody

SCR checkpoint engine

#2972 opened 2023-03-08 22:14 by adammoody

Codegen Inference Support

#2916 opened 2023-03-01 00:36 by RezaYazdaniAminabadi

pre/post forward calls to engine + generate method

#2832 opened 2023-02-14 21:17 by jeffra

Refactor/Pydantify autotuning config

#2769 opened 2023-01-31 01:54 by mrwyattii

Refactor/Pydantify main deepspeed config

#2757 opened 2023-01-28 01:34 by mrwyattii

Refactor/Pydantify profiling config

#2749 opened 2023-01-25 23:47 by mrwyattii

Refactor/Pydantify compression config

#2748 opened 2023-01-25 21:41 by mrwyattii

Preserve prior op builder backend api

#2705 opened 2023-01-14 01:40 by jeffra

Add correctness check for sharded checkpoint test

#2643 opened 2022-12-22 02:01 by mrwyattii

Refactor/Pydantify comms config

#2641 opened 2022-12-21 22:55 by mrwyattii

Refactor/Pydantify nebula checkpoint engine config

#2635 opened 2022-12-21 00:31 by mrwyattii

Refactor/Pydantify activation checkpointing config

#2633 opened 2022-12-20 23:21 by mrwyattii