[Iluvatar GPU] Modify the names of some variables (#3273)

Author: wuyujiji

docs/get_started/installation/iluvatar_gpu.md

@@ -32,7 +32,7 @@ docker exec -it paddle_infer bash
 ```bash
 pip3 install paddlepaddle==3.1.0a0 -i https://www.paddlepaddle.org.cn/packages/stable/cpu/
 pip3 install paddle-iluvatar-gpu==3.1.0 -i https://www.paddlepaddle.org.cn/packages/stable/ixuca/
-pip3 install fastdeploy_iluvatar_gpu -i https://www.paddlepaddle.org.cn/packages/stable/ixuca/ --extra-index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simplels
+pip3 install fastdeploy_iluvatar_gpu==2.1.0.dev0 -i https://www.paddlepaddle.org.cn/packages/stable/ixuca/ --extra-index-url https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simplels
 ```
 
 ## Prepare the inference demo script

fastdeploy/model_executor/ops/iluvatar/moe_ops.py

@@ -79,25 +79,27 @@ def group_gemm(
 def iluvatar_moe_expert_ffn(
     permute_input: paddle.Tensor,
     tokens_expert_prefix_sum: paddle.Tensor,
-    ffn1_weight: paddle.Tensor,
-    ffn2_weight: paddle.Tensor,
-    ffn1_bias: Optional[paddle.Tensor],
-    ffn1_scale: Optional[paddle.Tensor],
-    ffn2_scale: Optional[paddle.Tensor],
-    ffn2_in_scale: Optional[paddle.Tensor],
+    up_gate_proj_weight: paddle.Tensor,
+    down_proj_weight: paddle.Tensor,
+    up_gate_proj_bias: Optional[paddle.Tensor],
+    up_gate_proj_scale: Optional[paddle.Tensor],
+    down_proj_scale: Optional[paddle.Tensor],
+    down_proj_in_scale: Optional[paddle.Tensor],
     expert_idx_per_token: Optional[paddle.Tensor],
     quant_method: str,
     used_in_ep_low_latency: bool,
 ):
-    assert ffn1_bias is None
-    assert ffn1_scale is not None
-    assert ffn2_scale is not None
-    assert ffn2_in_scale is None
+    assert up_gate_proj_bias is None
+    assert up_gate_proj_scale is not None
+    assert down_proj_scale is not None
+    assert down_proj_in_scale is None
     assert expert_idx_per_token is None
     assert quant_method in ("weight_only_int8")
     assert not used_in_ep_low_latency
     tokens_expert_prefix_sum_cpu = tokens_expert_prefix_sum.to("cpu")
-    ffn1_output = w8a16_group_gemm(permute_input, ffn1_weight, ffn1_scale, tokens_expert_prefix_sum_cpu, -1)
+    ffn1_output = w8a16_group_gemm(
+        permute_input, up_gate_proj_weight, up_gate_proj_scale, tokens_expert_prefix_sum_cpu, -1
+    )
     act_out = swiglu(ffn1_output)
-    output = w8a16_group_gemm(act_out, ffn2_weight, ffn2_scale, tokens_expert_prefix_sum_cpu, -1)
+    output = w8a16_group_gemm(act_out, down_proj_weight, down_proj_scale, tokens_expert_prefix_sum_cpu, -1)
     return output

fastdeploy/worker/iluvatar_worker.py

@@ -88,6 +88,7 @@ def determine_available_memory(self) -> int:
         return int(float(os.getenv("FD_ILUVATAR_KVCACHE_MEM", "3")) * 1024**3)
 
 
+# TODO (yuzhe.wu): move it int work_process.py after baidu reconstructs the logic of workproc
 class IluvatarPaddleDisWorkerProc(PaddleDisWorkerProc):
     """
     Paddle Distributed wrapper for fastdeploy.worker.Worker,