github
/
ColossalAI
mirror of https://github.com/hpcaitech/ColossalAI
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
ColossalAI/extensions/csrc/cuda
History
yuehuayingxueluo 87079cffe8
[Inference]Support FP16/BF16 Flash Attention 2 And Add high_precision Flag To Rotary Embedding (#5461) 
* Support FP16/BF16 Flash Attention 2

* fix bugs in test_kv_cache_memcpy.py

* add context_kv_cache_memcpy_kernel.cu

* rm typename MT

* add tail process

* add high_precision

* add high_precision to config.py

* rm unused code

* change the comment for the high_precision parameter

* update test_rotary_embdding_unpad.py

* fix vector_copy_utils.h

* add comment for self.high_precision when using float32
 		2024-03-25 13:40:34 +08:00
..
include refactor code 2024-03-08 15:41:14 +08:00
pybind [Inference]Support FP16/BF16 Flash Attention 2 And Add high_precision Flag To Rotary Embedding (#5461) 2024-03-25 13:40:34 +08:00
utils [Inference]Support FP16/BF16 Flash Attention 2 And Add high_precision Flag To Rotary Embedding (#5461) 2024-03-25 13:40:34 +08:00
activation_kernel.cu add vec_type_trait implementation (#5473) 2024-03-19 18:36:40 +08:00
context_kv_cache_memcpy_kernel.cu [Inference]Support FP16/BF16 Flash Attention 2 And Add high_precision Flag To Rotary Embedding (#5461) 2024-03-25 13:40:34 +08:00
decode_kv_cache_memcpy_kernel.cu [Inference]Support FP16/BF16 Flash Attention 2 And Add high_precision Flag To Rotary Embedding (#5461) 2024-03-25 13:40:34 +08:00
fused_rotary_emb_and_cache_kernel.cu [Inference]Support FP16/BF16 Flash Attention 2 And Add high_precision Flag To Rotary Embedding (#5461) 2024-03-25 13:40:34 +08:00
layer_norm_kernel.cu refactor code 2024-03-11 17:06:57 +08:00
moe_kernel.cu refactor code 2024-03-11 17:06:57 +08:00
multi_tensor_adam_kernel.cu refactor code 2024-03-11 17:06:57 +08:00
multi_tensor_apply.cuh refactor code 2024-03-08 15:41:14 +08:00
multi_tensor_l2norm_kernel.cu refactor code 2024-03-08 15:41:14 +08:00
multi_tensor_lamb_kernel.cu refactor code 2024-03-11 17:06:57 +08:00
multi_tensor_scale_kernel.cu refactor code 2024-03-08 15:41:14 +08:00
multi_tensor_sgd_kernel.cu refactor code 2024-03-08 15:41:14 +08:00
rms_layernorm_kernel.cu refactor vector utils 2024-03-19 11:32:01 +08:00
scaled_masked_softmax.h refactor vector utils 2024-03-19 11:32:01 +08:00
scaled_masked_softmax_kernel.cu refactor code 2024-03-11 17:06:57 +08:00
scaled_upper_triang_masked_softmax.h [Inference]Support FP16/BF16 Flash Attention 2 And Add high_precision Flag To Rotary Embedding (#5461) 2024-03-25 13:40:34 +08:00
scaled_upper_triang_masked_softmax_kernel.cu refactor code 2024-03-11 17:06:57 +08:00