Merge branch 'main' into refactor-llm

requirements/runtime.txt

@@ -19,5 +19,8 @@ tiktoken
 torch<=2.1.2
 torchvision<=0.16.2
 # Minimum 4.36.0 to support `Cache` data structure used by KV Cache
-transformers>=4.36.0
+# Registering a causal mask in `LlamaModel` is not friendly for very large
+# `max_position_embeddings`. Refer to
+# https://github.com/huggingface/transformers/blob/v4.38.0/src/transformers/models/llama/modeling_llama.py#L921-L923
+transformers>=4.36.0,!=4.38.0,!=4.38.1,!=4.38.2
 transformers_stream_generator

xtuner/model/modules/dispatch/__init__.py

@@ -305,12 +305,12 @@ def dispatch_modules(model, use_varlen_attn=False):
         dispatch_internlm2_attn_forward(model, use_varlen_attn)
         if USE_TRITON_KERNEL:
             dispatch_internlm2_rmsnorm_forward(model)
-        # replace_internlm2_rote(model)
+        replace_internlm2_rote(model)
     elif 'internlm' in model_name:
         dispatch_internlm_attn_forward(model, use_varlen_attn)
         if USE_TRITON_KERNEL:
             dispatch_internlm_rmsnorm_forward(model)
-        # replace_internlm_rote(model)
+        replace_internlm_rote(model)
     elif 'llama' in model_name:
         dispatch_llama_attn_forward(model, use_varlen_attn)
         if USE_TRITON_KERNEL:

xtuner/model/modules/dispatch/llama.py

@@ -234,26 +234,18 @@ def llama_attn_forward_legacy(
     **kwargs,
 ) -> Tuple[torch.Tensor, Optional[torch.Tensor],
            Optional[Tuple[torch.Tensor]]]:
-    # LlamaFlashAttention2 attention does not support output_attentions
+    # Modified from https://github.com/huggingface/transformers/blob/ced9fd86f55ebb6b656c273f6e23f8ba50652f83/src/transformers/models/llama/modeling_llama.py#L331  # noqa:E501
     if 'padding_mask' in kwargs:
         warnings.warn(
-            'Passing `padding_mask` is deprecated and will be removed in v4.37'
-            ' Please make sure use `attention_mask` instead.`')
-
-        # overwrite attention_mask with padding_mask
-        attention_mask = kwargs.pop('padding_mask')
-
-    output_attentions = False
+            'Passing `padding_mask` is deprecated and will be removed in '
+            'v4.37. Please make sure use `attention_mask` instead.`')
 
     bsz, q_len, _ = hidden_states.size()
 
     query_states = self.q_proj(hidden_states)
     key_states = self.k_proj(hidden_states)
     value_states = self.v_proj(hidden_states)
 
-    # Flash attention requires the input to have the shape
-    # batch_size x seq_length x head_dim x hidden_dim
-    # therefore we just need to keep the original shape
     query_states = query_states.view(bsz, q_len, self.num_heads,
                                      self.head_dim).transpose(1, 2)
     key_states = key_states.view(bsz, q_len, self.num_key_value_heads,
@@ -263,6 +255,13 @@ def llama_attn_forward_legacy(
 
     kv_seq_len = key_states.shape[-2]
     if past_key_value is not None:
+        if self.layer_idx is None:
+            raise ValueError(
+                'The cache structure has changed since version v4.36. '
+                f'If you are using {self.__class__.__name__} '
+                'for auto-regressive decoding with k/v caching, '
+                'please make sure to initialize the attention class '
+                'with a layer index.')
         kv_seq_len += past_key_value.get_usable_length(kv_seq_len,
                                                        self.layer_idx)
     assert position_ids is not None
@@ -282,10 +281,6 @@ def llama_attn_forward_legacy(
     key_states = repeat_kv(key_states, self.num_key_value_groups)
     value_states = repeat_kv(value_states, self.num_key_value_groups)
 
-    # repeat kv for sequence parallel
-    key_states = repeat_kv_bshd(key_states, self.num_key_value_groups)
-    value_states = repeat_kv_bshd(value_states, self.num_key_value_groups)
-
     assert SUPPORT_FLASH2
     query_states = query_states.transpose(1, 2)
     key_states = key_states.transpose(1, 2)

xtuner/model/sft.py

@@ -206,13 +206,6 @@ def _build_from_cfg_or_module(self, cfg_or_mod):
             return cfg_or_mod
         elif isinstance(cfg_or_mod, dict):
             traverse_dict(cfg_or_mod)
-            if SUPPORT_FLASH2:
-                cfg_or_mod.torch_dtype = torch.bfloat16 \
-                    if torch.cuda.is_bf16_supported() else torch.float16
-                cfg_or_mod.attn_implementation = 'flash_attention_2'
-            if max_position_embeddings is not None:
-                cfg_or_mod = self._prepare_for_long_context_training(
-                    cfg_or_mod, max_position_embeddings)
             return BUILDER.build(cfg_or_mod)
         else:
             raise NotImplementedError