add npu support for baichuan

intel-analytics · Dec 19, 2024 · 9ab5879 · 9ab5879
1 parent 28e81fd
commit 9ab5879
Showing 1 changed file with 37 additions and 11 deletions.
diff --git a/python/llm/src/ipex_llm/transformers/npu_models/baichuan_mp.py b/python/llm/src/ipex_llm/transformers/npu_models/baichuan_mp.py
@@ -80,7 +80,8 @@ def __init__(
         intermediate_size,
         n_splits_linear: int = 1,
         n_splits_down_proj: int = 1,
-        group_size: int = 0
+        group_size: int = 0,
+        asym: bool = False,
     ):
         super().__init__(max_seq_len=max_seq_len,
                          transpose_value=transpose_value,
@@ -89,7 +90,8 @@ def __init__(
                          device=device,
                          n_splits_linear=n_splits_linear,
                          n_splits_down_proj=n_splits_down_proj,
-                         group_size=group_size)
+                         group_size=group_size,
+                         asym=asym)
         self.max_seq_len = max_seq_len
         self.intermediate_size = intermediate_size
         self.dtype = dtype
@@ -368,16 +370,19 @@ def __init__(
         do_print: bool = False,
         n_splits_linear: int = 1,
         n_splits_down_proj: int = 1,
-        group_size: int = 0
+        group_size: int = 0,
+        asym: bool = False,
     ):
         super().__init__()
 
         self.do_print = do_print
 
         op_parameters = []
         for w in parameters:
-            if isinstance(w, tuple):  # from QuantizedLinear
+            if isinstance(w, tuple) and not asym:  # from QuantizedLinear
                 op_parameters.append((w[0].numpy(), w[1].numpy()))
+            elif isinstance(w, tuple) and asym:  # from QuantizedLinear
+                op_parameters.append((w[0].numpy(), w[1].numpy(),  w[2].numpy()))
             elif w.dtype in [torch.int8, torch.uint8]:    # QuantizedLinear weight
                 op_parameters.append(w.numpy())
             elif isinstance(w, np.ndarray):     # scale
@@ -430,7 +435,8 @@ def __init__(
                 dtype=np_dtype,
                 n_splits_linear=n_splits_linear,
                 n_splits_down_proj=n_splits_down_proj,
-                group_size=group_size
+                group_size=group_size,
+                asym=asym,
             )
             self.backend_decoders.append(decoder)
 
@@ -506,7 +512,8 @@ def __init__(
         transpose_value: bool = False,
         n_splits_linear: int = 1,
         n_splits_down_proj: int = 1,
-        group_size: int = 0
+        group_size: int = 0,
+        asym: bool = False,
     ):
         super().__init__()
         self.op_parameters = parameters
@@ -537,7 +544,8 @@ def __init__(
             dtype=np_dtype,
             n_splits_linear=n_splits_linear,
             n_splits_down_proj=n_splits_down_proj,
-            group_size=group_size
+            group_size=group_size,
+            asym=asym
         )
         self.layer_norm_0 = layer_norm_0
         self.layer_norm_1 = layer_norm_1
@@ -620,6 +628,7 @@ def run_decode(
     layer_indexs = range(layer_start, layer_end)
     n_splits_linear = len(model.model.layers[0].mlp.gate_proj_dq_list)
     n_splits_down_proj = len(model.model.layers[0].mlp.down_proj_dq_list)
+    asym = getattr(model.config, "asym", False)
     for layer_idx in layer_indexs:
         curr_layer = model.model.layers[layer_idx]
         attn_layer = curr_layer.self_attn
@@ -631,10 +640,17 @@ def run_decode(
                            mlp_layer.down_proj_dq_list]:
             l_weights = []
             scales = []
+            zeros = []
             for l in layer_list:
                 l_weights.append(l.weight)
                 scales.append(l.scale)
-            weights.append((torch.stack(l_weights, axis=0), torch.stack(scales, axis=0)))
+                if l.zero is not None:
+                    zeros.append(l.zero)
+            if len(zeros):
+                weights.append((torch.stack(l_weights, axis=0), torch.stack(scales, axis=0),
+                                torch.stack(zeros, axis=0)))
+            else:
+                weights.append((torch.stack(l_weights, axis=0), torch.stack(scales, axis=0)))
 
         cached_cos = curr_layer.self_attn.rotary_emb.cos_cached.to(torch.float16)
         cached_sin = curr_layer.self_attn.rotary_emb.sin_cached.to(torch.float16)
@@ -663,7 +679,8 @@ def run_decode(
         do_print=False,
         n_splits_linear=n_splits_linear,
         n_splits_down_proj=n_splits_down_proj,
-        group_size=group_size
+        group_size=group_size,
+        asym=asym,
     )
 
     dist.barrier()
@@ -827,6 +844,7 @@ def run_prefill(
     layer_indexs = range(layer_start, layer_end)
     n_splits_linear = len(model.model.layers[0].mlp.gate_proj_dq_list)
     n_splits_down_proj = len(model.model.layers[0].mlp.down_proj_dq_list)
+    asym = getattr(model.config, "asym", False)
     for layer_idx in layer_indexs:
         curr_layer = model.model.layers[layer_idx]
         attn_layer = curr_layer.self_attn
@@ -838,10 +856,17 @@ def run_prefill(
                            mlp_layer.down_proj_dq_list]:
             l_weights = []
             scales = []
+            zeros = []
             for l in layer_list:
                 l_weights.append(l.weight)
                 scales.append(l.scale)
-            weights.append((torch.stack(l_weights, axis=0), torch.stack(scales, axis=0)))
+                if l.zero is not None:
+                    zeros.append(l.zero)
+            if len(zeros):
+                weights.append((torch.stack(l_weights, axis=0), torch.stack(scales, axis=0),
+                                torch.stack(zeros, axis=0)))
+            else:
+                weights.append((torch.stack(l_weights, axis=0), torch.stack(scales, axis=0)))
 
         cached_cos = curr_layer.self_attn.rotary_emb.cos_cached.to(torch.float16)
         cached_sin = curr_layer.self_attn.rotary_emb.sin_cached.to(torch.float16)
@@ -864,7 +889,8 @@ def run_prefill(
             transpose_value=transpose_value_cache,
             n_splits_linear=n_splits_linear,
             n_splits_down_proj=n_splits_down_proj,
-            group_size=group_size
+            group_size=group_size,
+            asym=asym
         )
 
         layer_weights.extend(weights)