Please enable JavaScript.
Coggle requires JavaScript to display documents.
MInference - Coggle Diagram
MInference
minference
patch.py
forward_llama_for_causal_lm()
forward_llama_model()
forward_llama_decoder_layer()
modules
minference_forward.py
minference_forward.forward
gather_last_q_vertical_slash_topk_v4
vertical_and_slash?
stream_llm?
block_sparse
dilated
vertical_and_slash_kernel
vertical_and_slash_kernel_extend
vertical_and_slash_kernel_static
dense
block_sparse_kernel
ops
pit_sparse_flash_attention_v2.py
vertical_slash_sparse_attention
csrc
vertical_slash_index.cu
convert_vertical_slash_indexes
convert_vertical_slash_indexes_64x64
convert_vertical_slash_indexes_kernel