Rex 563836496a

Refactor 2 awq gemm kernels into m16nXk32 (#2723) ...

Co-authored-by: Chunan Zeng <chunanzeng@Chunans-Air.attlocal.net>

2024-02-12 11:02:17 -08:00

..

layers

Refactor 2 awq gemm kernels into m16nXk32 (#2723)

2024-02-12 11:02:17 -08:00

models

Add fused top-K softmax kernel for MoE (#2769)

2024-02-05 17:38:02 -08:00

parallel_utils

[Minor] Fix false warning when TP=1 (#2674)

2024-01-30 14:39:40 -08:00

__init__.py

Refactor Worker & InputMetadata (#1843)

2023-11-29 22:16:37 -08:00

input_metadata.py

Support FP8-E5M2 KV Cache (#2279)

2024-01-28 16:43:54 -08:00

model_loader.py

Remove hardcoded device="cuda" to support more devices (#2503)

2024-02-01 15:46:39 -08:00

sampling_metadata.py

Use NCCL instead of ray for control-plane communication to remove serialization overhead (#2221)

2024-01-03 11:30:22 -08:00

utils.py

TP/quantization/weight loading refactor part 2 - Refactor quantized linear logic and extend quantization support to all models (#1622)

2023-11-15 22:50:41 -08:00

weight_utils.py

Use revision when downloading the quantization config file (#2697)

2024-02-01 15:41:58 -08:00