From 163fad337e7f0c315ae4573401fb941dab6b25e5 Mon Sep 17 00:00:00 2001 From: Zylan Date: Mon, 31 Mar 2025 14:33:36 +0800 Subject: [PATCH] support qwq-max --- app.py | 4 + config/models.json | 13 +++ models/__init__.py | 2 + models/alibaba.py | 253 ++++++++++++++++++++++++++++++++++++++++++ static/js/main.js | 53 +++++++++ static/js/settings.js | 1 + templates/index.html | 10 ++ 7 files changed, 336 insertions(+) create mode 100644 models/alibaba.py diff --git a/app.py b/app.py index 8cd3c46..f809956 100644 --- a/app.py +++ b/app.py @@ -81,6 +81,8 @@ def create_model_instance(model_id, api_keys, settings): api_key_id = "OpenaiApiKey" elif provider == 'deepseek': api_key_id = "DeepseekApiKey" + elif provider == 'alibaba': + api_key_id = "AlibabaApiKey" else: # 根据模型名称 if "claude" in model_id.lower(): @@ -89,6 +91,8 @@ def create_model_instance(model_id, api_keys, settings): api_key_id = "OpenaiApiKey" elif "deepseek" in model_id.lower(): api_key_id = "DeepseekApiKey" + elif "qvq" in model_id.lower() or "alibaba" in model_id.lower(): + api_key_id = "AlibabaApiKey" api_key = api_keys.get(api_key_id) if not api_key: diff --git a/config/models.json b/config/models.json index 5596372..117ac05 100644 --- a/config/models.json +++ b/config/models.json @@ -14,6 +14,11 @@ "name": "DeepSeek", "api_key_id": "DeepseekApiKey", "class_name": "DeepSeekModel" + }, + "alibaba": { + "name": "Alibaba", + "api_key_id": "AlibabaApiKey", + "class_name": "AlibabaModel" } }, "models": { @@ -56,6 +61,14 @@ "isReasoning": true, "version": "latest", "description": "DeepSeek推理模型，提供详细思考过程（仅支持文本）" + }, + "QVQ-Max-2025-03-25": { + "name": "QVQ-Max", + "provider": "alibaba", + "supportsMultimodal": true, + "isReasoning": true, + "version": "2025-03-25", + "description": "阿里巴巴通义千问-QVQ-Max版本，支持图像理解和思考过程" } } } \ No newline at end of file diff --git a/models/__init__.py b/models/__init__.py index eded06f..e3f4e65 100644 --- a/models/__init__.py +++ b/models/__init__.py @@ -2,6 +2,7 @@ from .base import BaseModel from .anthropic import AnthropicModel from .openai import OpenAIModel from .deepseek import DeepSeekModel +from .alibaba import AlibabaModel from .factory import ModelFactory __all__ = [ @@ -9,5 +10,6 @@ __all__ = [ 'AnthropicModel', 'OpenAIModel', 'DeepSeekModel', + 'AlibabaModel', 'ModelFactory' ] diff --git a/models/alibaba.py b/models/alibaba.py new file mode 100644 index 0000000..0e27d7d --- /dev/null +++ b/models/alibaba.py @@ -0,0 +1,253 @@ +import os +from typing import Generator, Dict, Optional, Any +from openai import OpenAI +from .base import BaseModel + +class AlibabaModel(BaseModel): + def get_default_system_prompt(self) -> str: + return """你是一位专业的问题分析与解答助手。当看到一个问题图片时，请： +1. 仔细阅读并理解问题 +2. 分析问题的关键组成部分 +3. 提供清晰的、逐步的解决方案 +4. 如果相关，解释涉及的概念或理论 +5. 如果有多种解决方法，先解释最高效的方法""" + + def get_model_identifier(self) -> str: + return "qvq-max" + + def analyze_text(self, text: str, proxies: dict = None) -> Generator[dict, None, None]: + """Stream QVQ-Max's response for text analysis""" + try: + # Initial status + yield {"status": "started", "content": ""} + + # Save original environment state + original_env = { + 'http_proxy': os.environ.get('http_proxy'), + 'https_proxy': os.environ.get('https_proxy') + } + + try: + # Set proxy environment variables if provided + if proxies: + if 'http' in proxies: + os.environ['http_proxy'] = proxies['http'] + if 'https' in proxies: + os.environ['https_proxy'] = proxies['https'] + + # Initialize OpenAI compatible client for DashScope + client = OpenAI( + api_key=self.api_key, + base_url="https://dashscope.aliyuncs.com/compatible-mode/v1" + ) + + # Prepare messages + messages = [ + { + "role": "system", + "content": [{"type": "text", "text": self.system_prompt}] + }, + { + "role": "user", + "content": [{"type": "text", "text": text}] + } + ] + + # 创建聊天完成请求 + response = client.chat.completions.create( + model=self.get_model_identifier(), + messages=messages, + temperature=self.temperature, + stream=True, + max_tokens=self.max_tokens if hasattr(self, 'max_tokens') and self.max_tokens else 4000 + ) + + # 记录思考过程和回答 + reasoning_content = "" + answer_content = "" + is_answering = False + + for chunk in response: + if not chunk.choices: + continue + + delta = chunk.choices[0].delta + + # 处理思考过程 + if hasattr(delta, 'reasoning_content') and delta.reasoning_content is not None: + reasoning_content += delta.reasoning_content + # 思考过程作为一个独立的内容发送 + yield { + "status": "reasoning", + "content": reasoning_content, + "is_reasoning": True + } + elif delta.content != "": + # 判断是否开始回答（从思考过程切换到回答） + if not is_answering: + is_answering = True + # 发送完整的思考过程 + if reasoning_content: + yield { + "status": "reasoning_complete", + "content": reasoning_content, + "is_reasoning": True + } + + # 累积回答内容 + answer_content += delta.content + + # 发送回答内容 + yield { + "status": "streaming", + "content": answer_content + } + + # 确保发送最终完整内容 + if answer_content: + yield { + "status": "completed", + "content": answer_content + } + + finally: + # Restore original environment state + for key, value in original_env.items(): + if value is None: + if key in os.environ: + del os.environ[key] + else: + os.environ[key] = value + + except Exception as e: + yield { + "status": "error", + "error": str(e) + } + + def analyze_image(self, image_data: str, proxies: dict = None) -> Generator[dict, None, None]: + """Stream QVQ-Max's response for image analysis""" + try: + # Initial status + yield {"status": "started", "content": ""} + + # Save original environment state + original_env = { + 'http_proxy': os.environ.get('http_proxy'), + 'https_proxy': os.environ.get('https_proxy') + } + + try: + # Set proxy environment variables if provided + if proxies: + if 'http' in proxies: + os.environ['http_proxy'] = proxies['http'] + if 'https' in proxies: + os.environ['https_proxy'] = proxies['https'] + + # Initialize OpenAI compatible client for DashScope + client = OpenAI( + api_key=self.api_key, + base_url="https://dashscope.aliyuncs.com/compatible-mode/v1" + ) + + # 检查系统提示词是否已包含语言设置指令 + system_prompt = self.system_prompt + language = self.language or '中文' + if not any(phrase in system_prompt for phrase in ['Please respond in', '请用', '使用', '回答']): + system_prompt = f"{system_prompt}\n\n请务必使用{language}回答，无论问题是什么语言。即使在分析图像时也请使用{language}回答。" + + # Prepare messages with image + messages = [ + { + "role": "system", + "content": [{"type": "text", "text": system_prompt}] + }, + { + "role": "user", + "content": [ + { + "type": "image_url", + "image_url": { + "url": f"data:image/jpeg;base64,{image_data}" + } + }, + { + "type": "text", + "text": "请分析这个图片并提供详细的解答。" + } + ] + } + ] + + # 创建聊天完成请求 + response = client.chat.completions.create( + model=self.get_model_identifier(), + messages=messages, + temperature=self.temperature, + stream=True, + max_tokens=self.max_tokens if hasattr(self, 'max_tokens') and self.max_tokens else 4000 + ) + + # 记录思考过程和回答 + reasoning_content = "" + answer_content = "" + is_answering = False + + for chunk in response: + if not chunk.choices: + continue + + delta = chunk.choices[0].delta + + # 处理思考过程 + if hasattr(delta, 'reasoning_content') and delta.reasoning_content is not None: + reasoning_content += delta.reasoning_content + # 思考过程作为一个独立的内容发送 + yield { + "status": "reasoning", + "content": reasoning_content, + "is_reasoning": True + } + elif delta.content != "": + # 判断是否开始回答（从思考过程切换到回答） + if not is_answering: + is_answering = True + # 发送完整的思考过程 + if reasoning_content: + yield { + "status": "reasoning_complete", + "content": reasoning_content, + "is_reasoning": True + } + + # 累积回答内容 + answer_content += delta.content + + # 发送回答内容 + yield { + "status": "streaming", + "content": answer_content + } + + # 确保发送最终完整内容 + if answer_content: + yield { + "status": "completed", + "content": answer_content + } + + finally: + # Restore original environment state + for key, value in original_env.items(): + if value is None: + if key in os.environ: + del os.environ[key] + else: + os.environ[key] = value + + except Exception as e: + yield { + "status": "error", + "error": str(e) + } \ No newline at end of file diff --git a/static/js/main.js b/static/js/main.js index a821e2a..fbb6502 100644 --- a/static/js/main.js +++ b/static/js/main.js @@ -355,6 +355,45 @@ class SnapSolver { } break; + case 'reasoning': + // 处理推理内容 (QVQ-Max模型使用) + if (data.content && this.thinkingContent && this.thinkingSection) { + console.log('Received reasoning content'); + this.thinkingSection.classList.remove('hidden'); + + // 记住用户的展开/折叠状态 + const wasExpanded = this.thinkingContent.classList.contains('expanded'); + + // 直接设置完整内容而不是追加 + this.setElementContent(this.thinkingContent, data.content); + + // 添加打字动画效果 + this.thinkingContent.classList.add('thinking-typing'); + + // 根据之前的状态决定是否展开 + if (wasExpanded) { + this.thinkingContent.classList.add('expanded'); + this.thinkingContent.classList.remove('collapsed'); + + // 更新切换按钮图标 + const toggleIcon = document.querySelector('#thinkingToggle .toggle-btn i'); + if (toggleIcon) { + toggleIcon.className = 'fas fa-chevron-up'; + } + } else { + // 初始状态为折叠 + this.thinkingContent.classList.add('collapsed'); + this.thinkingContent.classList.remove('expanded'); + + // 更新切换按钮图标 + const toggleIcon = document.querySelector('#thinkingToggle .toggle-btn i'); + if (toggleIcon) { + toggleIcon.className = 'fas fa-chevron-down'; + } + } + } + break; + case 'thinking_complete': // 完整的思考内容 if (data.content && this.thinkingContent && this.thinkingSection) { @@ -369,6 +408,20 @@ class SnapSolver { } break; + case 'reasoning_complete': + // 完整的推理内容 (QVQ-Max模型使用) + if (data.content && this.thinkingContent && this.thinkingSection) { + console.log('Reasoning complete'); + this.thinkingSection.classList.remove('hidden'); + + // 设置完整内容 + this.setElementContent(this.thinkingContent, data.content); + + // 移除打字动画 + this.thinkingContent.classList.remove('thinking-typing'); + } + break; + case 'streaming': if (data.content && this.responseContent) { console.log('Received content chunk'); diff --git a/static/js/settings.js b/static/js/settings.js index 032673e..b1e5fce 100644 --- a/static/js/settings.js +++ b/static/js/settings.js @@ -181,6 +181,7 @@ class SettingsManager { 'AnthropicApiKey': document.getElementById('AnthropicApiKey'), 'OpenaiApiKey': document.getElementById('OpenaiApiKey'), 'DeepseekApiKey': document.getElementById('DeepseekApiKey'), + 'AlibabaApiKey': document.getElementById('AlibabaApiKey'), 'mathpixAppId': this.mathpixAppIdInput, 'mathpixAppKey': this.mathpixAppKeyInput }; diff --git a/templates/index.html b/templates/index.html index ebbd208..668d0b3 100644 --- a/templates/index.html +++ b/templates/index.html @@ -7,6 +7,7 @@ + Snap Solver @@ -214,6 +215,15 @@ +
+ +
+ + +
+