[Inference] Fix bug in ChatGLM2 Tensor Parallelism (#5014)

* fix bug

* fix

* fix multiquery

* fix multiquery

---------

Co-authored-by: CjhHa1 <cjh18671720497outlook.com>

colossalai/inference/tensor_parallel/engine.py

@@ -77,14 +77,15 @@ class TPInferEngine:
         )
         self.layer_num = num_hidden_layers
 
-        self.multi_query_group_num = 0
+        self.multi_query_group_num = model.config.num_attention_heads
+        # default to attention_heads
+        self.multi_query_attention = model.config.multi_query_attention
 
         if hasattr(model.config, "multi_query_group_num"):
             self.multi_query_group_num = model.config.multi_query_group_num
 
         if hasattr(model.config, "num_key_value_heads"):
             self.multi_query_group_num = model.config.num_key_value_heads
-
         self.tp_size = -1  # to be set with given shard config in self.prepare_shard_config
         self.cache_manager = None
 
@@ -107,7 +108,7 @@ class TPInferEngine:
         assert self.head_num % self.tp_size == 0, f"Cannot shard {self.head_num} heads with tp size {self.tp_size}"
         self.head_num //= self.tp_size  # update sharded number of heads
 
-        if self.multi_query_group_num:
+        if self.multi_query_attention:
             # NOTE the logic of MQA tensor parallelism should be specified.
             assert (
                 self.multi_query_group_num % self.tp_size == 0

colossalai/inference/tensor_parallel/modeling/chatglm2.py

@@ -395,9 +395,9 @@ class ChatGLM2InferenceForwards:
         assert use_cache is True, "use_cache should be set to True using this chatglm attention"
         # hidden_states: original :[sq, b, h] --> this [b, sq, h]
         batch_size = hidden_states.shape[0]
+        hidden_size = hidden_states.shape[-1]
         # Attention heads [sq, b, h] --> [sq, b, (np * 3 * hn)]
         mixed_x_layer = self.query_key_value(hidden_states)
-
         if self.multi_query_attention:
             (query_layer, key_layer, value_layer) = mixed_x_layer.split(
                 [
@@ -437,7 +437,6 @@ class ChatGLM2InferenceForwards:
             mixed_x_layer = mixed_x_layer.view(*new_tensor_shape)
             # [sq, b, np, 3 * hn] --> 3 [sq, b, np, hn]
             (query_layer, key_layer, value_layer) = split_tensor_along_last_dim(mixed_x_layer, 3)
-
         cos, sin = infer_state.position_cos, infer_state.position_sin
 
         chatglm2_rotary_emb_fwd(
@@ -466,10 +465,10 @@ class ChatGLM2InferenceForwards:
         value_layer = value_layer.reshape(
             -1, self.num_multi_query_groups_per_partition, self.hidden_size_per_attention_head
         )
+
         if infer_state.is_context_stage:
             # first token generation:
             # copy key and value calculated in current step to memory manager
-
             copy_kv_to_mem_cache(
                 infer_state.decode_layer_id,
                 key_layer,
@@ -477,8 +476,7 @@ class ChatGLM2InferenceForwards:
                 infer_state.context_mem_index,
                 infer_state.cache_manager,
             )
-
-            attn_output = torch.empty_like(query_layer.view(-1, self.projection_size))
+            attn_output = torch.empty_like(query_layer.contiguous().view(-1, self.projection_size))
 
             # NOTE: no bug in context attn fwd (del it )
             lightllm_llama2_context_attention_fwd(
@@ -514,7 +512,7 @@ class ChatGLM2InferenceForwards:
                 )
 
             # second token and follows
-            attn_output = torch.empty_like(query_layer.view(-1, self.projection_size))
+            attn_output = torch.empty_like(query_layer.contiguous().view(-1, self.projection_size))
             cache_k = infer_state.cache_manager.key_buffer[infer_state.decode_layer_id][
                 : infer_state.decode_mem_end, :, :
             ]
@@ -542,6 +540,6 @@ class ChatGLM2InferenceForwards:
         # =================
         # Output:[b,sq, h]
         # =================
+        output = self.dense(attn_output).reshape(batch_size, -1, hidden_size)
 
-        output = self.dense(attn_output).reshape(batch_size, -1, self.projection_size)
         return output, kv_cache

colossalai/inference/tensor_parallel/policies/chatglm2.py

@@ -48,7 +48,10 @@ class ChatGLM2InferPolicy(ChatGLMModelPolicy):
         self.append_or_create_method_replacement(
             description=method_replacement, policy=policy, target_key=SelfAttention
         )
-
+        if self.shard_config.enable_tensor_parallelism:
+            policy[GLMBlock].attribute_replacement["self_attention.num_multi_query_groups_per_partition"] = (
+                self.model.config.multi_query_group_num // self.shard_config.tensor_parallel_size
+            )
         # for rmsnorm and others, we need to check the shape
         return policy
 

colossalai/shardformer/layer/linear.py

@@ -149,7 +149,6 @@ class Linear1D_Col(ParallelModule):
         out_features = module.out_features
         bias = module.bias is not None
         device = module.weight.device
-
         # ensure only one process group is passed
         if isinstance(process_group, (list, tuple)):
             assert len(process_group) == 1, f"Expected only one process group, got {len(process_group)}."

colossalai/shardformer/modeling/chatglm2_6b/modeling_chatglm.py

@@ -400,7 +400,6 @@ class SelfAttention(torch.nn.Module):
         )
 
         self.core_attention = CoreAttention(config, self.layer_number)
-
         # Output.
         self.dense = nn.Linear(
             self.projection_size,

colossalai/shardformer/policies/chatglm2.py

@@ -104,7 +104,6 @@ class ChatGLMPolicy(Policy):
                     ),
                 ],
             )
-
         # optimization configuration
         self.append_or_create_submodule_replacement(
             description=[

colossalai/shardformer/shard/sharder.py

@@ -180,7 +180,6 @@ class ModelSharder(object):
             assert target_module is not None, "target_module should not be None"
 
             native_sub_module = getattr_(org_layer, suffix, ignore=True)
-
             # Skip replacement if submodule is not kept by current device when pipeline parallel is enabled.
             if (include is not None) and (native_sub_module is not None) and (native_sub_module not in include):
                 continue