move state cache to memory (todo: state cache db)

README.md

@@ -73,6 +73,7 @@ body.json:
 - [x] CUDA operator int8 acceleration
 - [ ] macOS support
 - [ ] Linux support
+- [ ] Local State Cache DB
 
 ## Related Repositories:
 

README_ZH.md

@@ -73,6 +73,7 @@ body.json:
 - [x] CUDA算子int8提速
 - [ ] macOS支持
 - [ ] linux支持
+- [ ] 本地状态缓存数据库
 
 ## 相关仓库:
 

backend-python/routes/state_cache.py

@@ -3,6 +3,7 @@ from fastapi import APIRouter, HTTPException, Response, status
 from pydantic import BaseModel
 import gc
 import copy
+import torch
 
 router = APIRouter()
 
@@ -41,10 +42,14 @@ def add_state(body: AddStateBody):
         raise HTTPException(status.HTTP_400_BAD_REQUEST, "trie not loaded")
 
     id = trie.insert(body.prompt)
+    device = body.state[0].device
     dtrie[id] = {
         "tokens": copy.deepcopy(body.tokens),
-        "state": copy.deepcopy(body.state),
+        "state": [tensor.cpu() for tensor in body.state]
+        if device != torch.device("cpu")
+        else copy.deepcopy(body.state),
         "logits": copy.deepcopy(body.logits),
+        "device": device,
     }
 
     return "success"
@@ -77,11 +82,15 @@ def longest_prefix_state(body: LongestPrefixStateBody):
         pass
     if id != -1:
         v = dtrie[id]
+        device = v["device"]
         return {
             "prompt": trie[id],
             "tokens": v["tokens"],
-            "state": v["state"],
+            "state": [tensor.to(device) for tensor in v["state"]]
+            if device != torch.device("cpu")
+            else v["state"],
             "logits": v["logits"],
+            "device": device,
         }
     else:
         return {"prompt": "", "tokens": [], "state": None, "logits": None}