master

cann-robot[feat]support mxfp8 inference of GLM-5 on 950 platform

7c4fb613创建于 2 天前历史提交

文件	最后提交记录	最后更新时间
__init__.py	[Refactor] comm_manager refactor	23 天前
common_utils.py	[feat] mooncake + hixl, import, support models lazy import	25 天前
data_utils.py	[refactor] mc2 use independent communication group	1 个月前
forward_metadata.py	[refactor] add page attention cache management	1 个月前
graph_utils.py	[refactor] MTP refactor and R1 adaptation framework	2 个月前
hccl_utils.py	[feat]support mxfp8 inference of GLM-5 on 950 platform	2 天前
logging_config.py	refactor: 支持online多batch推理	1 个月前
profiler_context.py	[feat]prefill profiling	2 个月前
stream_utils.py	[feat]support mxfp8 inference of GLM-5 on 950 platform	2 天前