Add Pytorch Tensor Parallel support for Mistral (#34927)

add base tp support

src/transformers/models/mistral/configuration_mistral.py

@@ -97,6 +97,16 @@ class MistralConfig(PretrainedConfig):
 
     model_type = "mistral"
     keys_to_ignore_at_inference = ["past_key_values"]
+    # Default tensor parallel plan for base model `MistralModel`
+    base_model_tp_plan = {
+        "layers.*.self_attn.q_proj": "colwise",
+        "layers.*.self_attn.k_proj": "colwise",
+        "layers.*.self_attn.v_proj": "colwise",
+        "layers.*.self_attn.o_proj": "rowwise",
+        "layers.*.mlp.gate_proj": "colwise",
+        "layers.*.mlp.up_proj": "colwise",
+        "layers.*.mlp.down_proj": "rowwise",
+    }
 
     def __init__(
         self,

src/transformers/models/mistral/modeling_mistral.py

@@ -227,9 +227,9 @@ class MistralAttention(nn.Module):
         key_states = self.k_proj(hidden_states)
         value_states = self.v_proj(hidden_states)
 
-        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, self.head_dim).transpose(1, 2)
-        value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
+        query_states = query_states.view(bsz, q_len, -1, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(bsz, q_len, -1, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(bsz, q_len, -1, self.head_dim).transpose(1, 2)
 
         cos, sin = self.rotary_emb(value_states, position_ids)
         query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
@@ -983,6 +983,7 @@ class MistralModel(MistralPreTrainedModel):
 
 class MistralForCausalLM(MistralPreTrainedModel, GenerationMixin):
     _tied_weights_keys = ["lm_head.weight"]
+    _tp_plan = {"lm_head": "colwise_rep"}
 
     def __init__(self, config):
         super().__init__(config)