initial this repo

.gitattributes

@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* 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

deepseek-r1_tokenizer.py

@@ -0,0 +1,133 @@
+from transformers import AutoTokenizer, PreTrainedTokenizerFast
+from http.server import HTTPServer, BaseHTTPRequestHandler
+import json
+import argparse
+
+
+class Tokenizer_Http():
+
+    def __init__(self):
+        model_id = "deepseek-r1_tokenizer"
+        self.tokenizer = AutoTokenizer.from_pretrained(model_id)
+
+    def encode(self, prompt):
+        messages = [
+            {"role": "system", "content": "You are DeepSeek-R1, You are a helpful assistant."},
+            {"role": "user", "content": prompt}
+        ]
+        text = self.tokenizer.apply_chat_template(
+            messages,
+            tokenize=False,
+            add_generation_prompt=True
+        )
+        print(text)
+        token_ids = self.tokenizer.encode(text)
+        return token_ids
+
+    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
+
+
+tokenizer = Tokenizer_Http()
+
+print(tokenizer.bos_id, tokenizer.bos_token, tokenizer.eos_id, tokenizer.eos_token)
+print(tokenizer.encode("hello world"))
+
+
+class Request(BaseHTTPRequestHandler):
+    #通过类继承，新定义类
+    timeout = 5
+    server_version = 'Apache'
+
+    def do_GET(self):
+        print(self.path)
+        #在新类中定义get的内容（当客户端向该服务端使用get请求时，本服务端将如下运行）
+        self.send_response(200)
+        self.send_header("type", "get")  #设置响应头，可省略或设置多个
+        self.end_headers()
+
+        if self.path == '/bos_id':
+            bos_id = tokenizer.bos_id
+            # print(bos_id)
+            # to json
+            if bos_id is None:
+                msg = json.dumps({'bos_id': -1})
+            else:
+                msg = json.dumps({'bos_id': bos_id})
+        elif self.path == '/eos_id':
+            eos_id = tokenizer.eos_id
+            if eos_id is None:
+                msg = json.dumps({'eos_id': -1})
+            else:
+                msg = json.dumps({'eos_id': eos_id})
+        else:
+            msg = 'error'
+
+        print(msg)
+        msg = str(msg).encode()  #转为str再转为byte格式
+
+        self.wfile.write(msg)  #将byte格式的信息返回给客户端
+
+    def do_POST(self):
+        #在新类中定义post的内容（当客户端向该服务端使用post请求时，本服务端将如下运行）
+        data = self.rfile.read(int(
+            self.headers['content-length']))  #获取从客户端传入的参数（byte格式）
+        data = data.decode()  #将byte格式转为str格式
+
+        self.send_response(200)
+        self.send_header("type", "post")  #设置响应头，可省略或设置多个
+        self.end_headers()
+
+        if self.path == '/encode':
+            req = json.loads(data)
+            prompt = req['text']
+
+            token_ids = tokenizer.encode(prompt)
+            if token_ids is None:
+                msg = json.dumps({'token_ids': -1})
+            else:
+                msg = json.dumps({'token_ids': token_ids})
+
+        elif self.path == '/decode':
+            req = json.loads(data)
+            token_ids = req['token_ids']
+            text = tokenizer.decode(token_ids)
+            if text is None:
+                msg = json.dumps({'text': ""})
+            else:
+                msg = json.dumps({'text': text})
+        else:
+            msg = 'error'
+        print(msg)
+        msg = str(msg).encode()  #转为str再转为byte格式
+
+        self.wfile.write(msg)  #将byte格式的信息返回给客户端
+
+
+if __name__ == "__main__":
+
+    args = argparse.ArgumentParser()
+    args.add_argument('--host', type=str, default='localhost')
+    args.add_argument('--port', type=int, default=8080)
+    args = args.parse_args()
+
+    host = (args.host, args.port)  #设定地址与端口号，'localhost'等价于'127.0.0.1'
+    print('http://%s:%s' % host)
+    server = HTTPServer(host, Request)  #根据地址端口号和新定义的类，创建服务器实例
+    server.serve_forever()  #开启服务

deepseek-r1_tokenizer/tokenizer_config.json

@@ -0,0 +1,35 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<｜begin▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "<｜end▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "legacy": true,
+  "model_max_length": 16384,
+  "pad_token": {
+    "__type": "AddedToken",
+    "content": "<｜end▁of▁sentence｜>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sp_model_kwargs": {},
+  "unk_token": null,
+  "tokenizer_class": "LlamaTokenizerFast",
+  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% set ns = namespace(is_first=false, is_tool=false, is_output_first=true, system_prompt='') %}{%- for message in messages %}{%- if message['role'] == 'system' %}{% set ns.system_prompt = message['content'] %}{%- endif %}{%- endfor %}{{bos_token}}{{ns.system_prompt}}{%- for message in messages %}{%- if message['role'] == 'user' %}{%- set ns.is_tool = false -%}{{'<｜User｜>' + message['content']}}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is none %}{%- set ns.is_tool = false -%}{%- for tool in message['tool_calls']%}{%- if not ns.is_first %}{{'<｜Assistant｜><｜tool▁calls▁begin｜><｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{%- set ns.is_first = true -%}{%- else %}{{'\\n' + '<｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{{'<｜tool▁calls▁end｜><｜end▁of▁sentence｜>'}}{%- endif %}{%- endfor %}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is not none %}{%- if ns.is_tool %}{{'<｜tool▁outputs▁end｜>' + message['content'] + '<｜end▁of▁sentence｜>'}}{%- set ns.is_tool = false -%}{%- else %}{% set content = message['content'] %}{% if '</think>' in content %}{% set content = content.split('</think>')[-1] %}{% endif %}{{'<｜Assistant｜>' + content + '<｜end▁of▁sentence｜>'}}{%- endif %}{%- endif %}{%- if message['role'] == 'tool' %}{%- set ns.is_tool = true -%}{%- if ns.is_output_first %}{{'<｜tool▁outputs▁begin｜><｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- set ns.is_output_first = false %}{%- else %}{{'\\n<｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- endif %}{%- endif %}{%- endfor -%}{% if ns.is_tool %}{{'<｜tool▁outputs▁end｜>'}}{% endif %}{% if add_generation_prompt and not ns.is_tool %}{{'<｜Assistant｜>'}}{% endif %}"
+}

main_axcl_aarch64

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb111fc00c54abb6142a8f44df087bf104c8150a1cefa6be55c6b174b932c4ec
+size 999008

main_axcl_x86

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6cba0be8df523e351789fcfa745772423096f52a3b0a760f8d8c9f5b8bb2ec82
+size 1022384

main_prefill

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7661c63bcc2d2f6ca557ec81cad9371b1a196b0f9f171b6b91069e87a9ae9619
+size 3055912

post_config.json

@@ -0,0 +1,14 @@
+{
+    "enable_temperature" : true,
+    "temperature" : 0.9,
+    
+    "enable_repetition_penalty" : false,
+    "repetition_penalty" : 1.2,
+    "penalty_window" : 20,
+
+    "enable_top_p_sampling" : false,
+    "top_p" : 0.8,
+
+    "enable_top_k_sampling" : true,
+    "top_k" : 10
+}

run_deepseek-r1_1.5b_gptq_int4_ax650.sh

@@ -0,0 +1,14 @@
+./main_prefill \
+--template_filename_axmodel "deepseek-r1-1.5b-gptq-int4-ax650/qwen2_p128_l%d_together.axmodel" \
+--axmodel_num 28 \
+--tokenizer_type 2 \
+--filename_tokenizer_model "http://127.0.0.1:12345" \
+--bos 0 --eos 0 \
+--filename_post_axmodel "deepseek-r1-1.5b-gptq-int4-ax650/qwen2_post.axmodel" \
+--filename_tokens_embed "deepseek-r1-1.5b-gptq-int4-ax650/model.embed_tokens.weight.bfloat16.bin" \
+--tokens_embed_num 151936 \
+--tokens_embed_size 1536 \
+--use_mmap_load_embed 1 \
+--live_print 1 \
+--continue 1 \
+--prompt "$1"

run_deepseek-r1_1.5b_gptq_int4_axcl_aarch64.sh

@@ -0,0 +1,14 @@
+./main_axcl_aarch64 \
+--template_filename_axmodel "deepseek-r1-1.5b-gptq-int4-ax650/qwen2_p128_l%d_together.axmodel" \
+--axmodel_num 28 \
+--tokenizer_type 2 \
+--filename_tokenizer_model "http://127.0.0.1:12345" \
+--bos 0 --eos 0 \
+--filename_post_axmodel "deepseek-r1-1.5b-gptq-int4-ax650/qwen2_post.axmodel" \
+--filename_tokens_embed "deepseek-r1-1.5b-gptq-int4-ax650/model.embed_tokens.weight.bfloat16.bin" \
+--tokens_embed_num 151936 \
+--tokens_embed_size 1536 \
+--use_mmap_load_embed 0 \
+--live_print 1 \
+--continue 1 \
+--prompt "$1"

run_deepseek-r1_1.5b_gptq_int4_axcl_x86.sh

@@ -0,0 +1,14 @@
+./main_axcl_x86 \
+--template_filename_axmodel "deepseek-r1-1.5b-gptq-int4-ax650/qwen2_p128_l%d_together.axmodel" \
+--axmodel_num 28 \
+--tokenizer_type 2 \
+--filename_tokenizer_model "http://127.0.0.1:12345" \
+--bos 0 --eos 0 \
+--filename_post_axmodel "deepseek-r1-1.5b-gptq-int4-ax650/qwen2_post.axmodel" \
+--filename_tokens_embed "deepseek-r1-1.5b-gptq-int4-ax650/model.embed_tokens.weight.bfloat16.bin" \
+--tokens_embed_num 151936 \
+--tokens_embed_size 1536 \
+--use_mmap_load_embed 0 \
+--live_print 1 \
+--continue 1 \
+--prompt "$1"