compressor

产品支持情况

• Ascend 950PR/Ascend 950DT：支持

• Atlas A3 训练系列产品/Atlas A3 推理系列产品：支持

• Atlas A2 训练系列产品/Atlas A2 推理系列产品：不支持

• Atlas 200I/500 A2 推理产品：不支持

• Atlas 推理系列产品：不支持

• Atlas 训练系列产品：不支持

功能说明

• 接口功能：Compressor是推理场景下SMLA和QLI的前处理算子，用于将每4或128个token的KV cache压缩成一个，然后每个token与这些压缩的KV cache进行DSA计算。在长序列的情况下，Compressor可以有效地减少计算开销。主要计算过程为：

  1. 将输入$X$与$W^{KV}$做Matmul运算得到$kv\_state$，将输入$X$与$W^{Gate}$做Matmul运算后再与$Ape$做Add运算得到$score\_state$，$kv\_state$与$score\_state$根据输入的start_pos及cu_seqlens完成更新。
  2. 在coff为2的情况下对$kv\_state$和$score\_state$进行数据重排。
  3. 对$score\_state$进行softmax运算将softmax结果与$kv\_state$做Mul计算，后进行ReduceSum运算。

• 计算公式：

  1. 计算矩阵乘法：

  $$C4A：\left[kv\_stat{e}^a,score\_stat{e}^a\right]=X@\left[W^{aKV},W^{aGate}\right],\left[kv\_stat{e}^b,score\_stat{e}^b\right]=X@\left[W^{bKV},W^{bGate}\right];$$

  $$C128A：\left[kv\_state,score\_state\right]=X@\left[W^{KV},W^{Gate}\right]$$

  2. 计算分组加法：

  $$C4A：score\_stat{e}_i^{\prime }=\left[score\_stat{e}_{\left[4(i-1)+1:4i,:\right]}^a;score\_stat{e}_{\left[4i+1:4(i+1),:\right]}^b\right]+Ape,i=1,2,\cdots ,\frac{s}{4};$$

  $$C128A：score\_stat{e}_i^{\prime }=score\_stat{e}_{\left[128(i-1)+1:128i,:\right]}+Ape,i=1,2,\cdots ,\frac{s}{128};$$

  3. 计算分组Softmax：

  $$C4A：S_i^{\prime }=softmax(score\_stat{e}_i^{\prime }),i=1,2,\cdots ,\frac{s}{4};$$

  $$C128A：S_i^{\prime }=softmax(score\_stat{e}_i^{\prime }),i=1,2,\cdots ,\frac{s}{128};$$

  4. 计算Hadamard乘积：

  $$C4A：(S_H{)}_i=S_i^{\prime }\odot \left[kv\_stat{e}_{\left[4(i-1)+1:4i,:\right]}^a;kv\_stat{e}_{\left[4i+1:4(i+1),:\right]}^b\right],i=1,2,\cdots ,\frac{s}{4};$$

  $$C128A：S_H=S_i^{\prime }\odot kv\_state;$$

  5. 沿着压缩轴分组求和：

  $$C4A：C_i^{\text{Comp}}={\left[1\right]}_{1\times 8}@(S_H{)}_i,i=1,2,\cdots ,\frac{s}{4};$$

  $$C128A：C_i^{\text{Comp}}={\left[1\right]}_{1\times 128}@(S_H{)}_i,i=1,2,\cdots ,\frac{s}{128};$$

函数原型

cann_ops_transformer.compressor(
    x,
    wkv,
    wgate,
    state_cache,
    ape,
    cmp_ratio,
    *,
    state_block_table=None,
    cu_seqlens=None,
    seqused=None,
    start_pos=None,
    coff=1,
    cache_mode=1) -> Tensor

参数说明

参数名	参数类型	可选/必选	描述	数据类型	维度（shape）
x	Tensor	必选	原始不经压缩的数据，对应公式中的 $X$。不支持非连续，数据格式支持ND。	bfloat16、float16	[B,S,H]、[T,H]
wkv	Tensor	必选	kv压缩权重，对应公式中的 $W^{KV}$。不支持非连续，数据格式支持ND。	bfloat16、float16	[coff*D,H]
wgate	Tensor	必选	gate压缩权重，对应公式中的 $W^{Gate}$。不支持非连续，数据格式支持ND。	bfloat16、float16	[coff*D,H]
state_cache	Tensor	必选	kv_state和score_state的历史数据，对应公式中的 $\left[kv\_state,score\_state\right]$。不支持非连续，数据格式支持ND。	float32	[block_num, block_size, 2*coff*D]，要求block_num>0
ape	Tensor	必选	positional biases，对应公式中的 $Ape$。不支持非连续，数据格式支持ND。	float32	[cmp_ratio,coff*D]
cmp_ratio	int	必选	数据压缩率。默认值为4.	-	-
state_block_table	Tensor	可选	state_cache存储使用的block映射表。不支持非连续，数据格式支持ND。	int32	cache_mode=1时，shape为[B,ceil(Smax/block_size)]，Smax为每个Batch中最大的Sequence Length，当x的shape为[B,S,H]时，Smax=max(start_pos)+S。当x的shape为[T,H]时，Smax=max(start_pos)+max(cu_seqlens[n+1] - cu_seqlens[n])。cache_mode=2时，shape为[B]。当其中元素的值为0时，表示当前位置无需进行更新state_cache操作
cu_seqlens	Tensor	可选	不同Batch上的有效token数。不支持非连续，数据格式支持ND。 当x的shape为[B,S,H]时，参数必须为空。 当x的shape为[T,H]时，输入shape必须为[B+1,]，该参数为前缀和数组，后一个元素≥前一个元素，第一位必须为0。	int32	[B+1,]
seqused	Tensor	可选	不同Batch中实际参与压缩的token数。不支持非连续，数据格式支持ND。 指定为None时，数值等于每个Batch上的Sequence Length。 [B,S,H]场景：0 ≤ seqused[n] ≤ S [T,H]场景：0 ≤ seqused[n] ≤ cu_seqlens[n+1] - cu_seqlens[n]。	int32	[B,]
start_pos	Tensor	可选	计算起始位置。不支持非连续，数据格式支持ND，输入为None时从0开始计算	int32	[B,]
coff	int	可选	默认值1，仅支持1/2。 coff=1：无需进行overlap数据重排 coff=2：需要进行overlap数据重排。	int32	-
cache_mode	int	可选	state_cache的存储模式。 1：连续buffer 2：循环buffer 默认值1。	int32	-
state_cache_stride_dim0	int	可选	表示state_cache的0轴stride，默认值为0。	int32	-

• Atlas A3 训练系列产品/Atlas A3 推理系列产品：cache_mode不支持输入2，且不支持0轴非连续。

返回值说明

参数名	参数类型	可选/必选	描述	数据类型	维度（shape）
cmp_kv	Tensor	必选	压缩后的数据。不支持非连续，数据格式支持ND； 当x的shape为[B,S,H]时，输出拼接：(<batch0>compressed_tokens+pad0) + (<batch1>compressed_tokens+pad1) + ... + (<batchN>compressed_tokens+padN)； 当x的shape为[T,H]时，输出拼接：<batch0>compressed_tokens + <batch1>compressed_tokens + ... + <batchN>compressed_tokens + pad。	bfloat16、float16	x=[B,S,H]：[B,ceil(S/cmp_ratio),D] x=[T,H]：[min(T,T//cmp_ratio+B),D]

约束说明

• 该接口支持推理场景下使用。
• 该接口支持单算子模式和TorchAir图模式(aclgraph)调用。
• x参数维度含义：B（Batch Size）表示输入样本批量大小、S（Sequence Length）表示输入样本序列长度、H（Head Size）表示hidden层的大小、D（Head Dim）表示hidden层的最小单元大小、T表示所有Batch输入样本序列长度的累加和。
• 该接口支持B、S泛化，且存在如下场景限制：
  • 只支持B、S为0。
  • 部分长序列场景下，如果计算量过大可能会导致出现超过NPU内存的报错，注：这里计算量会受x输入shape的影响，值越大计算量越大。典型的长序列（即B、S的乘积或T较大）场景包括但不限于：

    B	S	H
    100	65525	4096
    25	261120	4096
    100	131072	4096
    100	261120	4096

• 支持D为128/512。
• 支持H为1K~10K，512对齐。
• 支持block_size为1~1024。
• 支持cmp_ratio为2/4/8/16/32/64/128。支持如下三种典型组合场景：
  • C4A: D=512, coff=2, cmp_ratio=4;
  • C4Li: D=128, coff=2, cmp_ratio=4;
  • C128A: D=512, coff=1, cmp_ratio=128。

确定性计算

• 默认支持确定性计算。

调用说明

• 单算子模式调用：

  import torch
  import torch_npu
  import numpy as np
  from cann_ops_transformer.ops import compressor
  import torch.nn as nn
  import math
  
  def get_seq_used_by_batch(batch_idx, S, seqused, cu_seqlens):
      if seqused is not None:
          return seqused[batch_idx]
      else:
          if cu_seqlens is not None:
              return cu_seqlens[batch_idx + 1] - cu_seqlens[batch_idx]
          else:
              return S
  
  data_type = torch.bfloat16
  hidden_size = 4096
  coff = 1 # 1:no overlap 2:overlap
  cmp_ratio = 128
  cache_mode = 1
  head_dim = 512
  cu_seqlens = [0, 1]
  # -------------
  B = 1
  S = 1
  S_max = 0
  block_size = 128
  start_pos = [8191] * B # (B,)
  start_p=8191
  seqused = None # (B,), None时cu_seqlens的数据全部参与计算，否则按传参实际值计算
  
  # BS是否合轴
  bs_combine_flag = True
  update_flag = 1
  save_state_seqlens = None
  if seqused is not None:
      seqused = torch.tensor(seqused).to(torch.int32)
  if start_pos is not None:
      start_pos = torch.tensor(start_pos).to(torch.int32)
  else:
      start_pos = torch.full((B,), start_p, dtype=torch.int32)
  
  if bs_combine_flag:
      if cu_seqlens is None:
          T = B * S
          if T !=0:
              cu_seqlens = torch.arange(0, T + 1, S, dtype=torch.int32)
          else:
              cu_seqlens = torch.zeros((B+1), dtype=torch.int32)
      else:
          cu_seqlens = torch.tensor(cu_seqlens).to(torch.int32)
      for i in range(B):
          if start_pos[i] + cu_seqlens[i + 1] - cu_seqlens[i] > S_max:
              S_max = start_pos[i] + cu_seqlens[i + 1] - cu_seqlens[i]
  else:
      cu_seqlens = None
      S_max = max(start_pos) + S
  ### ======================== gen input data start =============================
  # page state
  if cache_mode == 1:
      max_block_num_per_batch = (S_max + block_size - 1) // block_size
      block_num = B * max_block_num_per_batch
      next_block_id = 1
      print(f"max_block_num_per_batch: {max_block_num_per_batch}")
      block_table = torch.zeros(size=(B, max_block_num_per_batch), dtype=torch.int32)
      for i in range(B):
          # 需要读取state的范围
          cur_start = start_pos[i] // cmp_ratio * cmp_ratio - cmp_ratio
          cur_end = start_pos[i] // cmp_ratio * cmp_ratio + cmp_ratio
          if start_pos[i] % cmp_ratio == 0:
              cur_end = start_pos[i]
          cur_end = min(cur_end, start_pos[i] + S)
          cur_start_block_id = (cur_start // block_size) if cur_start >= 0 else 0
          cur_end_block_id = (cur_end - 1) // block_size
          for j in range(cur_start_block_id, cur_end_block_id + 1):
              block_table[i][j] = next_block_id
              next_block_id = next_block_id + 1
          # 需要写入state的范围
          end_pos = get_seq_used_by_batch(i, S, seqused, cu_seqlens)
          if save_state_seqlens is not None:
              next_start = start_pos[i] + end_pos - save_state_seqlens[i]
              next_end = start_pos[i] + end_pos
          else:
              next_start = (start_pos[i] + end_pos) // cmp_ratio * cmp_ratio - cmp_ratio
              next_end = (start_pos[i] + end_pos) // cmp_ratio * cmp_ratio + cmp_ratio
              if (start_pos[i] + end_pos) % cmp_ratio == 0:
                  next_end = start_pos[i] + end_pos
          next_end = min(next_end, start_pos[i] + end_pos)
          next_start_block_id = (next_start // block_size) if next_start >= 0 else 0
          next_end_block_id = (next_end - 1) // block_size
          for j in range(next_start_block_id, next_end_block_id + 1):
              if block_table[i][j] == 0:
                  block_table[i][j] = next_block_id
                  next_block_id = next_block_id + 1
  
      if B==0:
          kv_state = torch.tensor(np.random.uniform(-10, 10, (0, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(-10, 10, (0, block_size, coff * head_dim))).to(torch.float32)
      else:
          kv_state = torch.tensor(np.random.uniform(-10, 10, (torch.max(block_table) + 1, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(-10, 10, (torch.max(block_table) + 1, block_size, coff * head_dim))).to(torch.float32)
  else:
      block_table = torch.tensor(random.sample(list(range(B)), B), dtype=torch.int32)
      block_size = (2 * cmp_ratio + S - 1) if coff == 2 else (cmp_ratio + S - 1)
      if B==0:
          kv_state = torch.tensor(np.random.uniform(kv_state_datarange[0], kv_state_datarange[1], (0, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(score_state_datarange[0], score_state_datarange[1], (0, block_size, coff * head_dim))).to(torch.float32)
      else:
          kv_state = torch.tensor(np.random.uniform(kv_state_datarange[0], kv_state_datarange[1], (B, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(score_state_datarange[0], score_state_datarange[1], (B, block_size, coff * head_dim))).to(torch.float32)
  
  # other input
  if bs_combine_flag:
      x_shape = (cu_seqlens[-1], hidden_size)
  else:
      x_shape = (B, S, hidden_size)
  
  x = torch.tensor(np.random.uniform(-10.0, 10.0, x_shape)).to(data_type).npu()
  wkv = torch.tensor(np.random.uniform(-10, 10, (coff * head_dim, hidden_size))).to(data_type).npu()
  wgate = torch.tensor(np.random.uniform(-10, 10, (coff * head_dim, hidden_size))).to(data_type).npu()
  ape = torch.tensor(np.random.uniform(-10, 10, (cmp_ratio, coff * head_dim))).to(torch.float32).npu()
  if cache_mode == 1:  # 连续buffer
      state_cache = torch.zeros((kv_state.shape[0], kv_state.shape[1], 2*kv_state.shape[2]))
      state_cache = state_cache.npu()
      state_cache[:, :, :state_cache.shape[2]//2] = kv_state.clone()
      state_cache[:, :, state_cache.shape[2]//2:] = score_state.clone()
  else:
      layer_pad = random.randint(1, 50)
      layer_start_idx = random.randint(0, layer_pad-1)
      print(f"layer_pad: {layer_pad}")
      print(f"layer_start_idx: {layer_start_idx}")
      state_cache_pad = torch.zeros((kv_state.shape[0],kv_state.shape[1]*kv_state.shape[2]*2+layer_pad))
      print(f"state_cache_pad: shape {state_cache_pad.shape}")
      state_cache_pad = state_cache_pad.to("npu:%s" % DEVICE_ID)
      state_cache = state_cache_pad[:, layer_start_idx : layer_start_idx + kv_state.shape[1]*kv_state.shape[2]*2].view(-1, kv_state.shape[1], kv_state.shape[2]*2)
      state_cache = state_cache.to("npu:%s" % DEVICE_ID)
      state_cache[:, :, :state_cache.shape[2]//2] = kv_state.clone()
      state_cache[:, :, state_cache.shape[2]//2:] = score_state.clone()
      print(f"state_cache: shape {state_cache.shape}, dtype: {state_cache.dtype}, is_contiguous: {state_cache.is_contiguous()}, stride0: {state_cache.stride(0)}")
  
  block_table = block_table.npu()
  start_pos = torch.tensor(start_pos).to(torch.int32).npu()
  if cu_seqlens is not None:
      cu_seqlens = torch.tensor(cu_seqlens).to(torch.int32).npu()
  if seqused is not None:
      seqused = torch.tensor(seqused).to(torch.int32).npu()
  
  cmp_kv = (
      compressor(
          x,
          wkv,
          wgate,
          state_cache,
          ape,
          cmp_ratio = cmp_ratio,
          state_block_table = block_table,
          cu_seqlens = cu_seqlens,
          seqused = seqused,
          start_pos = start_pos,
          coff = coff,
          cache_mode = cache_mode
      )
  )
  

• TorchAir图模式调用：

  import torch
  import torch_npu
  import numpy as np
  import torch.nn as nn
  import torchair
  from cann_ops_transformer.ops import compressor
  import math
  
  def get_seq_used_by_batch(batch_idx, S, seqused, cu_seqlens):
      if seqused is not None:
          return seqused[batch_idx]
      else:
          if cu_seqlens is not None:
              return cu_seqlens[batch_idx + 1] - cu_seqlens[batch_idx]
          else:
              return S
  
  data_type = torch.bfloat16
  hidden_size = 4096
  coff = 1 # 1:no overlap 2:overlap
  cmp_ratio = 128
  cache_mode = 1
  head_dim = 512
  cu_seqlens = [0, 1]
  # -------------
  B = 1
  S = 1
  S_max = 0
  block_size = 128
  start_pos = [8191] * B # (B,)
  start_p=8191
  seqused = None # (B,), None时cu_seqlens的数据全部参与计算，否则按传参实际值计算
  
  # BS是否合轴
  bs_combine_flag = True
  update_flag = 1
  save_state_seqlens = None
  if seqused is not None:
      seqused = torch.tensor(seqused).to(torch.int32)
  if start_pos is not None:
      start_pos = torch.tensor(start_pos).to(torch.int32)
  else:
      start_pos = torch.full((B,), start_p, dtype=torch.int32)
  
  if bs_combine_flag:
      if cu_seqlens is None:
          T = B * S
          if T !=0:
              cu_seqlens = torch.arange(0, T + 1, S, dtype=torch.int32)
          else:
              cu_seqlens = torch.zeros((B+1), dtype=torch.int32)
      else:
          cu_seqlens = torch.tensor(cu_seqlens).to(torch.int32)
      for i in range(B):
          if start_pos[i] + cu_seqlens[i + 1] - cu_seqlens[i] > S_max:
              S_max = start_pos[i] + cu_seqlens[i + 1] - cu_seqlens[i]
  else:
      cu_seqlens = None
      S_max = max(start_pos) + S
  ### ======================== gen input data start =============================
  # page state
  if cache_mode == 1:
      max_block_num_per_batch = (S_max + block_size - 1) // block_size
      block_num = B * max_block_num_per_batch
      next_block_id = 1
      print(f"max_block_num_per_batch: {max_block_num_per_batch}")
      block_table = torch.zeros(size=(B, max_block_num_per_batch), dtype=torch.int32)
      for i in range(B):
          # 需要读取state的范围
          cur_start = start_pos[i] // cmp_ratio * cmp_ratio - cmp_ratio
          cur_end = start_pos[i] // cmp_ratio * cmp_ratio + cmp_ratio
          if start_pos[i] % cmp_ratio == 0:
              cur_end = start_pos[i]
          cur_end = min(cur_end, start_pos[i] + S)
          cur_start_block_id = (cur_start // block_size) if cur_start >= 0 else 0
          cur_end_block_id = (cur_end - 1) // block_size
          for j in range(cur_start_block_id, cur_end_block_id + 1):
              block_table[i][j] = next_block_id
              next_block_id = next_block_id + 1
          # 需要写入state的范围
          end_pos = get_seq_used_by_batch(i, S, seqused, cu_seqlens)
          if save_state_seqlens is not None:
              next_start = start_pos[i] + end_pos - save_state_seqlens[i]
              next_end = start_pos[i] + end_pos
          else:
              next_start = (start_pos[i] + end_pos) // cmp_ratio * cmp_ratio - cmp_ratio
              next_end = (start_pos[i] + end_pos) // cmp_ratio * cmp_ratio + cmp_ratio
              if (start_pos[i] + end_pos) % cmp_ratio == 0:
                  next_end = start_pos[i] + end_pos
          next_end = min(next_end, start_pos[i] + end_pos)
          next_start_block_id = (next_start // block_size) if next_start >= 0 else 0
          next_end_block_id = (next_end - 1) // block_size
          for j in range(next_start_block_id, next_end_block_id + 1):
              if block_table[i][j] == 0:
                  block_table[i][j] = next_block_id
                  next_block_id = next_block_id + 1
  
      if B==0:
          kv_state = torch.tensor(np.random.uniform(-10, 10, (0, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(-10, 10, (0, block_size, coff * head_dim))).to(torch.float32)
      else:
          kv_state = torch.tensor(np.random.uniform(-10, 10, (torch.max(block_table) + 1, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(-10, 10, (torch.max(block_table) + 1, block_size, coff * head_dim))).to(torch.float32)
  else:
      block_table = torch.tensor(random.sample(list(range(B)), B), dtype=torch.int32)
      block_size = (2 * cmp_ratio + S - 1) if coff == 2 else (cmp_ratio + S - 1)
      if B==0:
          kv_state = torch.tensor(np.random.uniform(kv_state_datarange[0], kv_state_datarange[1], (0, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(score_state_datarange[0], score_state_datarange[1], (0, block_size, coff * head_dim))).to(torch.float32)
      else:
          kv_state = torch.tensor(np.random.uniform(kv_state_datarange[0], kv_state_datarange[1], (B, block_size, coff * head_dim))).to(torch.float32)
          score_state = torch.tensor(np.random.uniform(score_state_datarange[0], score_state_datarange[1], (B, block_size, coff * head_dim))).to(torch.float32)
  
  # other input
  if bs_combine_flag:
      x_shape = (cu_seqlens[-1], hidden_size)
  else:
      x_shape = (B, S, hidden_size)
  
  x = torch.tensor(np.random.uniform(-10.0, 10.0, x_shape)).to(data_type).npu()
  wkv = torch.tensor(np.random.uniform(-10, 10, (coff * head_dim, hidden_size))).to(data_type).npu()
  wgate = torch.tensor(np.random.uniform(-10, 10, (coff * head_dim, hidden_size))).to(data_type).npu()
  ape = torch.tensor(np.random.uniform(-10, 10, (cmp_ratio, coff * head_dim))).to(torch.float32).npu()
  if cache_mode == 1:  # 连续buffer
      state_cache = torch.zeros((kv_state.shape[0], kv_state.shape[1], 2*kv_state.shape[2]))
      state_cache = state_cache.npu()
      state_cache[:, :, :state_cache.shape[2]//2] = kv_state.clone()
      state_cache[:, :, state_cache.shape[2]//2:] = score_state.clone()
  else:
      layer_pad = random.randint(1, 50)
      layer_start_idx = random.randint(0, layer_pad-1)
      print(f"layer_pad: {layer_pad}")
      print(f"layer_start_idx: {layer_start_idx}")
      state_cache_pad = torch.zeros((kv_state.shape[0],kv_state.shape[1]*kv_state.shape[2]*2+layer_pad))
      print(f"state_cache_pad: shape {state_cache_pad.shape}")
      state_cache_pad = state_cache_pad.to("npu:%s" % DEVICE_ID)
      state_cache = state_cache_pad[:, layer_start_idx : layer_start_idx + kv_state.shape[1]*kv_state.shape[2]*2].view(-1, kv_state.shape[1], kv_state.shape[2]*2)
      state_cache = state_cache.to("npu:%s" % DEVICE_ID)
      state_cache[:, :, :state_cache.shape[2]//2] = kv_state.clone()
      state_cache[:, :, state_cache.shape[2]//2:] = score_state.clone()
      print(f"state_cache: shape {state_cache.shape}, dtype: {state_cache.dtype}, is_contiguous: {state_cache.is_contiguous()}, stride0: {state_cache.stride(0)}")
  
  block_table = block_table.npu()
  start_pos = torch.tensor(start_pos).to(torch.int32).npu()
  if cu_seqlens is not None:
      cu_seqlens = torch.tensor(cu_seqlens).to(torch.int32).npu()
  if seqused is not None:
      seqused = torch.tensor(seqused).to(torch.int32).npu()
  
  class CompressorNetwork(nn.Module):
      def __init__(self):
          super(CompressorNetwork, self).__init__()
  
      def forward(self, x, wkv, wgate, state_cache, ape, cmp_ratio, state_block_table = None, cu_seqlens = None,
                  seqused = None, start_pos = None, coff = 1, cache_mode = 1):
          cmp_kv = (
              torch.ops.cann_ops_transformer.compressor(
                  x,
                  wkv,
                  wgate,
                  state_cache,
                  ape,
                  cmp_ratio = cmp_ratio,
                  state_block_table = state_block_table,
                  cu_seqlens = cu_seqlens,
                  seqused = seqused,
                  start_pos = start_pos,
                  coff = coff,
                  cache_mode = cache_mode
              )
          )
          return cmp_kv
  
  from torchair.configs.compiler_config import CompilerConfig
  config = CompilerConfig()
  config.mode = "reduce-overhead"
  npu_backend = torchair.get_npu_backend(compiler_config=config)
  torch._dynamo.reset()
  npu_mode = torch.compile(CompressorNetwork(), fullgraph=True, backend=npu_backend, dynamic=False)
  cmp_kv = npu_mode(
                  x,
                  wkv,
                  wgate,
                  state_cache,
                  ape,
                  cmp_ratio = cmp_ratio,
                  state_block_table = block_table,
                  cu_seqlens = cu_seqlens,
                  seqused = seqused,
                  start_pos = start_pos,
                  coff = coff,
                  cache_mode = cache_mode)