↓ 1 callersMethod_call(a: torch.Tensor, b: torch.Tensor, c: torch.Tensor, bias: torch.Tensor, P: torch.Tensor,
locks:
analysis/baselines/gemm/triton_gemm_v02.py:25
↓ 1 callersMethod_call(a: torch.Tensor, b: torch.Tensor, c: torch.Tensor, bias: torch.Tensor, P: torch.Tensor,
locks:
analysis/baselines/gemm/utils/gemm_wrapper.py:21
↓ 1 callersFunction_update_kv_cachekv: (batch_size, seqlen, 2, nheads, head_dim) or (batch_size, 1, 2, nheads, head_dim)
training/llama/llama/models/mha.py:16
↓ 1 callersMethod_update_kv_cachekv: (batch_size, seqlen, 2, nheads, head_dim) or (batch_size, 1, 2, nheads, head_dim)
training/llama/llama/models/mha.py:159
↓ 1 callersFunctionapply_rotary_emb_kv_ Arguments: kv: (batch_size, seqlen, 2, nheads, headdim) cos, sin: (seqlen, rotary_dim / 2) interleaved: if True, rotate p
training/llama/llama/models/rotary.py:308
↓ 1 callersFunctionapply_rotary_emb_qkv_ Arguments: qkv: (batch_size, seqlen, 3, nheads, headdim) or (batch_size, seqlen, num_heads_q + 2 * num_heads_k, headdim). If
training/llama/llama/models/rotary.py:236