Upload folder using huggingface_hub

.gitattributes

@@ -37,3 +37,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 main_axcl_aarch64 filter=lfs diff=lfs merge=lfs -text
 main_axcl_x86 filter=lfs diff=lfs merge=lfs -text
 main_prefill filter=lfs diff=lfs merge=lfs -text
+main_ax650 filter=lfs diff=lfs merge=lfs -text

main_ax650

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b44029895e2d61f23b2eb9761de5fe1387b76c36937b2daef86b048538e8b7f4
+size 1035576

main_axcl_x86

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6cba0be8df523e351789fcfa745772423096f52a3b0a760f8d8c9f5b8bb2ec82
-size 1022384
+oid sha256:928d36be31c15d081a7d346464f41458e9624d8b68d5f7dfb3d3189686ce2754
+size 8421624

run_smollm2_360m_ax650.sh

@@ -1,14 +1,10 @@
-./main_prefill \
+./main_ax650 \
 --template_filename_axmodel "smollm2-360m-ax650/llama_p128_l%d_together.axmodel" \
 --axmodel_num 32 \
---tokenizer_type 2 \
---filename_tokenizer_model "http://127.0.0.1:12345" \
---bos 0 --eos 0 \
+--url_tokenizer_model "http://127.0.0.1:12345" \
 --filename_post_axmodel "smollm2-360m-ax650/llama_post.axmodel" \
 --filename_tokens_embed "smollm2-360m-ax650/model.embed_tokens.weight.bfloat16.bin" \
 --tokens_embed_num 49152 \
 --tokens_embed_size 960 \
 --use_mmap_load_embed 1 \
---live_print 1 \
---continue 1 \
---prompt "$1"
+--live_print 1

run_smollm2_360m_axcl_x86.sh

@@ -1,14 +1,11 @@
 ./main_axcl_x86 \
 --template_filename_axmodel "smollm2-360m-ax650/llama_p128_l%d_together.axmodel" \
 --axmodel_num 32 \
---tokenizer_type 2 \
---filename_tokenizer_model "http://127.0.0.1:12345" \
---bos 0 --eos 0 \
+--url_tokenizer_model "http://127.0.0.1:12345" \
 --filename_post_axmodel "smollm2-360m-ax650/llama_post.axmodel" \
 --filename_tokens_embed "smollm2-360m-ax650/model.embed_tokens.weight.bfloat16.bin" \
 --tokens_embed_num 49152 \
 --tokens_embed_size 960 \
 --use_mmap_load_embed 0 \
 --live_print 1 \
---continue 1 \
---prompt "$1"
+--devices 0

smollm2-360m-ax650/llama_p128_l0_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:489febc320a376e049b6fa6f059dcf423174513da6544b89b36e89cbf8900fd7
-size 11218690
+oid sha256:c0734291ef438bc620885f6ce352691bb44b2b8cfa3debbb2b7e2288eddeac0e
+size 14366662

smollm2-360m-ax650/llama_p128_l10_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:676db9c8def427b8d24e89154d7c54623ad96d550a74b93392af5b2351609dc2
-size 11218690
+oid sha256:aa4a2f9399164c1696d5971b1561a801bb7ce427341fdb8239474a7682067977
+size 14366662

smollm2-360m-ax650/llama_p128_l11_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a572f7f626eb436047c9fdcc14e12f448faebaa7ac7ec54710e25f8283a2ae91
-size 11218690
+oid sha256:8444eeabb0a2a5f0be583c82ca144c8769ce589cd0ba6a55be8cebfaba28e48a
+size 14366662

smollm2-360m-ax650/llama_p128_l12_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19aac1ab81b32be3d2b6c32bb30b59157d48e85ece6c2aac36099ea371082b35
-size 11218690
+oid sha256:606aa0cfcb409c5e3c0cf2a6f22137efa4445084383db5da0bacb575738e44c9
+size 14366662

smollm2-360m-ax650/llama_p128_l13_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8d1ae96d861cafeed478b4953b0a17f018bf2d3d7b65704509ec4462f0d931ba
-size 11218690
+oid sha256:80371853a9ce6c2fb9337aaedd351746d730eb2053aa21284245dde374e93325
+size 14366662

smollm2-360m-ax650/llama_p128_l14_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aaeefd23274a3aab6018781683b03c097fd07d565c779b9c01cf65e30d42801f
-size 11218690
+oid sha256:58c8fd27bb7790d44af4b2e99561c93b1b3b2a6f785e356da808939d3cdcca49
+size 14366662

smollm2-360m-ax650/llama_p128_l15_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:84bd88ec4ce38d35ca28996c1b178953b3ea2d97eb4a1a6941ed2a4a61b678fc
-size 11218690
+oid sha256:860d69bd418ed76270b9500de9bf437fd15b4b0ed699c52aaefc444e692e97e6
+size 14366662

smollm2-360m-ax650/llama_p128_l16_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cf68c070d51d404f1c1c861e8e5d9e7e7c3421ee59a7790f30050a056933c62e
-size 11218690
+oid sha256:5c77d3184833d4b26d55307cc373790e9f73a4c7b212bdcda3400cf3081eee34
+size 14366662

smollm2-360m-ax650/llama_p128_l17_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:90dabb01513932459c2c344f8aee8f8c53d6869b6a5ab91251ef18fd139a619c
-size 11218690
+oid sha256:8baae17086192c85503dda9579be16f7f4c2a7da35b70472d70083436e83d2b5
+size 14366662

smollm2-360m-ax650/llama_p128_l18_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f5ed70f209cfb6859971bfc6acc289e6768fd160799a54a4a30e0bcf7860bb05
-size 11218690
+oid sha256:42318a0d6236662ffd3fb8d4f8270d3f0baf07ab744b6654ac59be36af1fb360
+size 14366662

smollm2-360m-ax650/llama_p128_l19_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:13d65f6805ed97aea4c813b7bcf90821ec0ad6ce7e0a1be0d385554cb8211c34
-size 11218690
+oid sha256:ed1647460528bc5f5505802fd8ea650d5c0402681b64a7a8b5b17d7ec0e040c1
+size 14366662

smollm2-360m-ax650/llama_p128_l1_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:995f165341d9808fa11f754236041555edf933571aec9d9938bf480ac8f05328
-size 11218690
+oid sha256:f5387f73421834ed85bd21bc0fdd2101bf3f8114d5b39d2dfa395f724b23f6c1
+size 14366662

smollm2-360m-ax650/llama_p128_l20_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7da1158e22b5ca6a3332949d1c456a62574cd0ce274b0c247b685856f3329f31
-size 11218690
+oid sha256:bad98fc8bef9a2cf1465b73ed362cfc6de39600bc96a793870f05e974a3afbe3
+size 14366662

smollm2-360m-ax650/llama_p128_l21_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3f2ae999d342e4a76292a0117f9604187c8985cd07cb2ef11e4f57c7381bcc0a
-size 11218690
+oid sha256:ca5b70b89d707e9f8c842dd76bda089a5094eaea7a86b9cf3274d40e5651a0d5
+size 14366662

smollm2-360m-ax650/llama_p128_l22_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d1b428789458de91e3fb4cd95d6e58c400e06795dbcc3cffea1e78c0d58d902a
-size 11218690
+oid sha256:a954432716f4198be22bc96cf12f2fc1c2832c3dc29ca3c9cfef4f65d653d697
+size 14366662

smollm2-360m-ax650/llama_p128_l23_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad53076c95797aebf892595a2b7d88fb3e712e27efdfd92a54561c750a88124d
-size 11218690
+oid sha256:588404cd3596ac507e08ab638af14b35b392c31724c4f16da182360cd24443ef
+size 14366662

smollm2-360m-ax650/llama_p128_l24_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0bcf51f105a596b88a8d06f8095cad551b337a4384afb645fb554d0e346a7382
-size 11218690
+oid sha256:0aa585194294c5769f362abeb9790a3886385552870c1f7be94a60f9d534f343
+size 14366662

smollm2-360m-ax650/llama_p128_l25_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:002e44afa0542f838bf9fde6d217ac92c6855c5bfc9c8802e8e80c5fb0509687
-size 11218690
+oid sha256:f04d3c96ed6d97d6b63de777108a28903ae6a0717edb3cfab60454a19451c92e
+size 14366662

smollm2-360m-ax650/llama_p128_l26_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7aa19ca187a16f2f08afc80ff9f755e78cf18868f45b699c065c2bb1fb194290
-size 11218690
+oid sha256:1803820151e703fe04cd66de507a67a873139450883633bf27d9b664866cd55d
+size 14366662

smollm2-360m-ax650/llama_p128_l27_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:da4bf6b5fdc6abd69f7cc6bd259df7685972de282f07e66def6236a209623820
-size 11218690
+oid sha256:af87c16a7b1237bd8a5c4c2e6060b37e28ad7e1827f07cfbecc8e9fe47eb45fa
+size 14366662

smollm2-360m-ax650/llama_p128_l28_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:416d24cf85eb4eab1cb5bc19cea300318b4a0d34b63ccf591f5264ef827e9bfa
-size 11218690
+oid sha256:e1cfdfd1a672c94883e30090a055b79bfdffc7fc9a35a42a561e8c0969256258
+size 14366662

smollm2-360m-ax650/llama_p128_l29_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:579e1b979e97247aed78dabe564d50b5bdbc81c0edada4046c81c43cd76b645e
-size 11218690
+oid sha256:fbc1631ac57bf0be7ba8427addafc57a028606cd247542102e59c2958ed0fb1e
+size 14366662

smollm2-360m-ax650/llama_p128_l2_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fabbd458282bb1dc52006a1c96c68fcdcb7c8b686a557a1da08dc15fc29e4b80
-size 11218690
+oid sha256:27e02a84be7684c319a25aac8c79bb764ae7d7b3c14016f8a097f8892e5560cd
+size 14366662

smollm2-360m-ax650/llama_p128_l30_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f7bd3399a0a851f05116de727c0a28838dace9f8096876fe782cf8383fe314a5
-size 11218690
+oid sha256:d887968d0c5d5e72800bb9b6f5c4e8f04cfa6ee459c1305a4ceaf78dd51b88d9
+size 14366662

smollm2-360m-ax650/llama_p128_l31_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f5078029f054ba1c4a2ecf0dd0919865084017b39d8c790049227bdffb1d048c
-size 11218690
+oid sha256:a1ecd507b28a797923162b90c7a6f8692cd2723768111c2e8d3ca626b7e24364
+size 14366662

smollm2-360m-ax650/llama_p128_l3_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0983d31036313e367a7cf91bb771bee3cb1854e07baf4068bff123ffd296c98e
-size 11218690
+oid sha256:103590a50f2d3f980f45af873fd4967549a4fc0cf6e586c8bcad6813a9d25e4b
+size 14366662

smollm2-360m-ax650/llama_p128_l4_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:14b9402ae0d8f05d08b525fc571f641f219138e92f2ce9b82ac89cb6ffc15910
-size 11218690
+oid sha256:4ec8f24b916c26e46b53c3d203e41ee77653a80995eb60fbe3908c0233dc4e5c
+size 14366662

smollm2-360m-ax650/llama_p128_l5_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4d2143709587c18dce172eb8297fa7ff57bdf397a1a0c4f3e9fd56d63720dc2a
-size 11218690
+oid sha256:74173cde61d2d9418ffd83e89c61e9a50b206f96cfc7a5e8c9a1a85a73ae4acf
+size 14366662

smollm2-360m-ax650/llama_p128_l6_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff12125e92ed34c5bfe5efd7232229d69b23956248236e34d3357c711a78736a
-size 11218690
+oid sha256:9b8ebc7f64fbd5222ff55fcb2597d511934b483e4ff32bdf402b56d9bf99a557
+size 14366662

smollm2-360m-ax650/llama_p128_l7_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5eb6216f8f4531364348e7e85403e305103eef19ac7ba18b33b7fff0afdc2d9e
-size 11218690
+oid sha256:3bb9c44849bb310da63a8e5a7a700f5518aead0527ce867a8752d3cce307d9c0
+size 14366662

smollm2-360m-ax650/llama_p128_l8_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af8cebf4cf6698ebfdb07663a18cc9bc046ae52944f51902315a3a7df72c318c
-size 11218690
+oid sha256:c220ec7bfd85fa83c078750d0f413a14b91d1b942288cf4f8e818b74f634339b
+size 14366662

smollm2-360m-ax650/llama_p128_l9_together.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:33b7bbe08d59ee7d2c187edf8c85cee8aee464fe11fb1c5608a44369fb83907c
-size 11218690
+oid sha256:11b1b2937e45983377fdf7ce6d0f628bf11855815143b5d986e783a6fb6f0936
+size 14366662

smollm2-360m-ax650/llama_post.axmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f891122c5016004037af8b7dd70edd218729548301e432425d0e9c2305610adb
-size 51443158
+oid sha256:4de8c64b8b547a60d0835065dd233540e29afcd07558b7c228a028ef7751cbb8
+size 51443106

smollm2-360m-ax650/model.embed_tokens.weight.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a914ff7a25ddaba760cbff63fff38f95f5ca9cea3fecf6c766050fb752bd829e
+size 188743808

smollm2_tokenizer/chat_template.jinja

@@ -0,0 +1,6 @@
+{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system
+You are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>
+' }}{% endif %}{{'<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>' + '
+'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}

smollm2_tokenizer/special_tokens_map.json

@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>"
+  ],
+  "bos_token": {
+    "content": "<|im_start|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

smollm2_tokenizer/tokenizer_config.json

@@ -143,9 +143,9 @@
     "<|im_end|>"
   ],
   "bos_token": "<|im_start|>",
-  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful AI assistant named SmolLM, trained by Hugging Face<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
   "clean_up_tokenization_spaces": false,
   "eos_token": "<|im_end|>",
+  "extra_special_tokens": {},
   "model_max_length": 8192,
   "pad_token": "<|im_end|>",
   "tokenizer_class": "GPT2Tokenizer",

smollm2_tokenizer_uid.py

@@ -0,0 +1,187 @@
+from transformers import AutoTokenizer, PreTrainedTokenizerFast
+from http.server import HTTPServer, BaseHTTPRequestHandler
+import json
+import argparse
+import uuid
+
+# 全局字典：存储 uid 到 Tokenizer_Http 实例的映射
+tokenizers = {}
+
+class Tokenizer_Http():
+    def __init__(self):
+        model_id = "smollm2_tokenizer"
+        self.tokenizer = AutoTokenizer.from_pretrained(model_id)
+        self.messages = [{"role": "system", "content": "I are a AI assistant, created by HuggingfaceTB"}]
+        self.token_ids = []
+
+    def encode(self, prompt, last_reply=None):
+        if last_reply is not None:
+            self.messages.append({"role": "assistant", "content": last_reply})
+            text = self.tokenizer.apply_chat_template(
+                self.messages,
+                tokenize=False,
+                add_generation_prompt=True
+            )
+            # print("生成的文本:\n============\n", text, "============\n")
+            self.token_ids = self.tokenizer.encode(text)[:-3]
+        self.messages.append({"role": "user", "content": prompt})
+        
+        text = self.tokenizer.apply_chat_template(
+            self.messages,
+            tokenize=False,
+            add_generation_prompt=True
+        )
+        print("生成的文本:\n============\n", text, "============\n")
+        token_ids = self.tokenizer.encode(text)
+        # 找出新增部分
+        diff = token_ids[len(self.token_ids):]
+        self.token_ids = token_ids
+        print(self.decode(diff))
+        return token_ids, diff
+
+    def decode(self, token_ids):
+        return self.tokenizer.decode(token_ids)
+
+    @property
+    def bos_id(self):
+        return self.tokenizer.bos_token_id
+
+    @property
+    def eos_id(self):
+        return self.tokenizer.eos_token_id
+    
+    @property
+    def bos_token(self):
+        return self.tokenizer.bos_token
+
+    @property
+    def eos_token(self):
+        return self.tokenizer.eos_token
+    
+    def reset(self, system_prompt="I are a AI assistant, created by HuggingfaceTB"):
+        self.messages = [
+            {"role": "system", "content": system_prompt},
+        ]
+        text = self.tokenizer.apply_chat_template(
+            self.messages,
+            tokenize=False,
+            add_generation_prompt=True
+        )
+        token_ids = self.tokenizer.encode(text)[:-3]
+        self.token_ids = token_ids
+        print(self.decode(token_ids))
+        return token_ids
+        
+
+class Request(BaseHTTPRequestHandler):
+    timeout = 5
+    server_version = 'Apache'
+
+    def do_GET(self):
+        print("GET 请求路径:", self.path)
+        self.send_response(200)
+        self.send_header("Content-Type", "application/json")
+        self.end_headers()
+
+        # 新增接口：获取 uid
+        if '/get_uid' in self.path:
+            new_uid = str(uuid.uuid4())
+            print("新 uid:", new_uid)
+            # 为该 uid 创建一个新的 Tokenizer_Http 实例
+            tokenizers[new_uid] = Tokenizer_Http()
+            msg = json.dumps({'uid': new_uid})
+        elif '/bos_id' in self.path:
+            # 获取 uid 参数（例如 ?uid=xxx）
+            uid = self.get_query_param("uid")
+            instance: Tokenizer_Http = tokenizers.get(uid)
+            if instance is None:
+                msg = json.dumps({'error': 'Invalid uid'})
+            else:
+                bos_id = instance.bos_id
+                msg = json.dumps({'bos_id': bos_id if bos_id is not None else -1})
+        elif '/eos_id' in self.path:
+            uid = self.get_query_param("uid")
+            instance: Tokenizer_Http = tokenizers.get(uid)
+            if instance is None:
+                msg = json.dumps({'error': 'Invalid uid'})
+            else:
+                eos_id = instance.eos_id
+                msg = json.dumps({'eos_id': eos_id if eos_id is not None else -1})
+        else:
+            msg = json.dumps({'error': 'Invalid GET endpoint'})
+
+        print("响应消息:", msg)
+        self.wfile.write(msg.encode())
+
+    def do_POST(self):
+        content_length = int(self.headers.get('content-length', 0))
+        data = self.rfile.read(content_length).decode()
+        print("POST 请求路径:", self.path)
+        print("接收到的数据:", data)
+        req = json.loads(data)
+
+        self.send_response(200)
+        self.send_header("Content-Type", "application/json")
+        self.end_headers()
+
+        if '/encode' in self.path:
+            # 请求数据中必须包含 uid, text, 和可选的 last_reply
+            uid = req.get('uid')
+            prompt = req.get('text')
+            last_reply = req.get('last_reply')
+            instance: Tokenizer_Http = tokenizers.get(uid)
+            if instance is None:
+                msg = json.dumps({'error': 'Invalid uid'})
+            else:
+                token_ids, diff = instance.encode(prompt, last_reply)
+                msg = json.dumps({'token_ids': token_ids, 'diff': diff})
+        elif '/decode' in self.path:
+            uid = req.get('uid')
+            token_ids = req.get('token_ids')
+            instance: Tokenizer_Http = tokenizers.get(uid)
+            if instance is None:
+                msg = json.dumps({'error': 'Invalid uid'})
+            else:
+                text = instance.decode(token_ids)
+                msg = json.dumps({'text': text})
+        elif '/reset' in self.path:
+            uid = req.get("uid")
+            system_prompt = req.get("system_prompt")
+            instance: Tokenizer_Http = tokenizers.get(uid)
+            if instance is None:
+                msg = json.dumps({'error': 'Invalid uid'})
+            else:
+                if system_prompt is not None:
+                    print("system_prompt:", system_prompt)
+                    token_ids = instance.reset(system_prompt)
+                    msg = json.dumps({'token_ids': token_ids})
+                else:
+                    token_ids = instance.reset()
+                    msg = json.dumps({'token_ids': token_ids})
+        else:
+            msg = json.dumps({'error': 'Invalid POST endpoint'})
+
+        print("响应消息:", msg)
+        self.wfile.write(msg.encode())
+
+    def get_query_param(self, key):
+        """
+        辅助函数：从 GET 请求的 URL 中获取查询参数的值
+        例如：/bos_id?uid=xxx
+        """
+        from urllib.parse import urlparse, parse_qs
+        query = urlparse(self.path).query
+        params = parse_qs(query)
+        values = params.get(key)
+        return values[0] if values else None
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--host', type=str, default='0.0.0.0')
+    parser.add_argument('--port', type=int, default=12345)
+    args = parser.parse_args()
+
+    host = (args.host, args.port)
+    print('Server running at http://%s:%s' % host)
+    server = HTTPServer(host, Request)
+    server.serve_forever()