done

README.md

@@ -235,6 +235,18 @@ git clone https://github.com/jingyaogong/minimind.git
 pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple
 ```
 
+因为需要的库较多，建议使用虚拟环境来安装：
+```bash
+python -m venv .venv
+```
+
+再下载库：
+
+```bash
+pip install -r requirements.txt
+```
+
+
 ### 2.下载模型
 到项目根目录
 ```bash
@@ -243,11 +255,21 @@ git clone https://huggingface.co/jingyaogong/MiniMind2
 
 ### （可选）命令行问答
 
+如果在第1步中安装了虚拟环境，需要在使用命令行激活虚拟环境：
+```bash
+.venv/Scripts/activate
+```
+
 ```bash
 # load=0: load from pytorch model, load=1: load from transformers-hf model
 python eval_model.py --load 1 --model_mode 2
 ```
 
+或者使用
+```bash
+python model_chat.py
+```
+
 ### （可选）启动WebUI
 
 ```bash

model_chat.py

@@ -1,21 +1,17 @@
+print("开始加载模型")
+
 import argparse
 import random
-import warnings
 import numpy as np
-from transformers import AutoTokenizer, AutoModelForCausalLM, TextStreamer
+from transformers import AutoTokenizer, TextStreamer, LlamaForCausalLM
-from model.model_minimind import MiniMindConfig, MiniMindForCausalLM
+import torch
-from model.model_lora import *
 
-warnings.filterwarnings('ignore')
+def init_model():
-
-
-def init_model(args):
     transformers_model_path = './MiniMind2'
     tokenizer = AutoTokenizer.from_pretrained(transformers_model_path)
-    model = AutoModelForCausalLM.from_pretrained(transformers_model_path, trust_remote_code=True)
+    model = LlamaForCausalLM.from_pretrained(transformers_model_path, trust_remote_code=True)
-    print(f'MiniMind模型参数量: {sum(p.numel() for p in model.parameters() if p.requires_grad) / 1e6:.2f}M(illion)')
+    print(f'模型参数量：{sum(p.numel() for p in model.parameters() if p.requires_grad) / 1e6:.2f}M(illion)')
-    return model.eval().to(args.device), tokenizer
+    return model.eval().to('cuda' if torch.cuda.is_available() else 'cpu'), tokenizer
-
 
 # 设置可复现的随机种子
 def setup_seed(seed):
@@ -27,74 +23,57 @@ def setup_seed(seed):
     torch.backends.cudnn.deterministic = True
     torch.backends.cudnn.benchmark = False
 
+parser = argparse.ArgumentParser(description="Chat with MiniMind")
+parser.add_argument('--temperature', default=0.85, type=float)
+parser.add_argument('--top_p', default=0.85, type=float)
 
-def main():
+parser.add_argument('--max_seq_len', default=8192, type=int)
-    print("开始加载模型")
+args = parser.parse_args()
-    parser = argparse.ArgumentParser(description="Chat with MiniMind")
+model, tokenizer = init_model()
-    parser.add_argument('--lora_name', default='None', type=str)
-    parser.add_argument('--out_dir', default='out', type=str)
-    parser.add_argument('--temperature', default=0.85, type=float)
-    parser.add_argument('--top_p', default=0.85, type=float)
-    parser.add_argument('--device', default='cuda' if torch.cuda.is_available() else 'cpu', type=str)
-    # 此处max_seq_len（最大输出长度）并不意味模型具有对应的长文本的性能，仅防止QA出现被截断的问题
-    # MiniMind2-moe (145M)：(hidden_size=640, num_hidden_layers=8, use_moe=True)
-    # MiniMind2-Small (26M)：(hidden_size=512, num_hidden_layers=8)
-    # MiniMind2 (104M)：(hidden_size=768, num_hidden_layers=16)
-    parser.add_argument('--hidden_size', default=512, type=int)
-    parser.add_argument('--num_hidden_layers', default=8, type=int)
-    parser.add_argument('--max_seq_len', default=8192, type=int)
-    parser.add_argument('--use_moe', default=False, type=bool)
-    # 携带历史对话上下文条数
-    # history_cnt需要设为偶数，即【用户问题, 模型回答】为1组；设置为0时，即当前query不携带历史上文
-    # 模型未经过外推微调时，在更长的上下文的chat_template时难免出现性能的明显退化，因此需要注意此处设置
-    parser.add_argument('--history_cnt', default=0, type=int)
-    parser.add_argument('--load', default=0, type=int, help="0: 原生torch权重，1: transformers加载")
-    parser.add_argument('--model_mode', default=1, type=int,
-                        help="0: 预训练模型，1: SFT-Chat模型，2: RLHF-Chat模型，3: Reason模型，4: RLAIF-Chat模型")
-    args = parser.parse_args()
 
-    model, tokenizer = init_model(args)
+streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
 
-    streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
+print("模型加载完毕")
 
-    messages = []
+messages = []
-    for prompt in iter(lambda: input('👶: '), ''):
+for prompt in iter(lambda: input('👶: '), ''):
-        if prompt == "exit":
+    if prompt == "exit":
-            break
+        break
-        setup_seed(random.randint(0, 2048))
 
-        messages.append({"role": "user", "content": prompt})
+    elif print == "clear history":
+        messages = []
+        print("成功清除对话历史")
+        continue
+    setup_seed(random.randint(0, 2048))
 
-        new_prompt = tokenizer.apply_chat_template(
+    messages.append({"role": "user", "content": prompt})
-            messages,
-            tokenize=False,
-            add_generation_prompt=True
-        ) if args.model_mode != 0 else (tokenizer.bos_token + prompt)
 
-        inputs = tokenizer(
+    new_prompt = tokenizer.apply_chat_template(
-            new_prompt,
+        messages,
-            return_tensors="pt",
+        tokenize=False,
-            truncation=True
+        add_generation_prompt=True
-        ).to(args.device)
+    )
 
-        print('🤖️: ', end='')
+    inputs = tokenizer(
-        generated_ids = model.generate(
+        new_prompt,
-            inputs["input_ids"],
+        return_tensors="pt",
-            max_new_tokens=args.max_seq_len,
+        truncation=True
-            num_return_sequences=1,
+    ).to('cuda' if torch.cuda.is_available() else 'cpu')
-            do_sample=True,
-            attention_mask=inputs["attention_mask"],
-            pad_token_id=tokenizer.pad_token_id,
-            eos_token_id=tokenizer.eos_token_id,
-            streamer=streamer,
-            top_p=args.top_p,
-            temperature=args.temperature
-        )
 
-        response = tokenizer.decode(generated_ids[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
+    print('🤖️: ', end='')
-        messages.append({"role": "assistant", "content": response})
+    generated_ids = model.generate(
-        print('\n')
+        inputs["input_ids"],
+        max_new_tokens=args.max_seq_len,
+        num_return_sequences=1,
+        do_sample=True,
+        attention_mask=inputs["attention_mask"],
+        pad_token_id=tokenizer.pad_token_id,
+        eos_token_id=tokenizer.eos_token_id,
+        streamer=streamer,
+        top_p=args.top_p,
+        temperature=args.temperature
+    )
 
-
+    response = tokenizer.decode(generated_ids[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
-if __name__ == "__main__":
+    messages.append({"role": "assistant", "content": response})
-    main()
+    print('\n')