can't work with deepseek-reasoner

[uuip]

2025-05-06 13:35:59 INFO  Ollama Server - 500 Internal Server Error: POST - /api/chat in 187ms
2025-05-06 13:35:59 ERROR Ollama Server - Error processing request.
io.ktor.client.plugins.ClientRequestException: Client request(POST https://api.deepseek.com/chat/completions) invalid: 400 Bad Request. Text: "{"error":{"message":"deepseek-reasoner does not support successive user or assistant messages (messages[0] and messages[1] in your input). You should interleave the user/assistant messages in the message sequence.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}"
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invokeSuspend(DefaultResponseValidation.kt:54)
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invoke(DefaultResponseValidation.kt)
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invoke(DefaultResponseValidation.kt)
	at io.ktor.client.plugins.HttpCallValidatorKt.HttpCallValidator$lambda$2$validateResponse(HttpCallValidator.kt:110)
	at io.ktor.client.plugins.HttpCallValidatorKt.access$HttpCallValidator$lambda$2$validateResponse(HttpCallValidator.kt:1)
	at io.ktor.client.plugins.HttpCallValidatorKt$HttpCallValidator$2$2.invokeSuspend(HttpCallValidator.kt:129)
	at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:33)
	at kotlinx.coroutines.DispatchedTask.run(DispatchedTask.kt:100)
	at kotlinx.coroutines.internal.LimitedDispatcher$Worker.run(LimitedDispatcher.kt:113)
	at kotlinx.coroutines.scheduling.TaskImpl.run(Tasks.kt:89)
	at kotlinx.coroutines.scheduling.CoroutineScheduler.runSafely(CoroutineScheduler.kt:586)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.executeTask(CoroutineScheduler.kt:820)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.runWorker(CoroutineScheduler.kt:717)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.run(CoroutineScheduler.kt:704)

Can't fetch any result from deepseek when uploading a file from Jetbrain IDE AI Chat plugin.

[Stream29]

Can you provide more background to help me to reproduct this issue? So I can find the reason.

This doesn't seem to be an issue with the Proxy; it's most likely a problem on DeepSeek's official side. Just moments ago, when I used the chat on DeepSeek's official website, it showed a server error or something similar—though it's now back to normal. Also, thanks to the author for providing the plugin!

[uuip]

2025-05-06 14:26:56 INFO  Config - Config file watcher started at /Users/sharp/Desktop/ai_prj/ProxyAsLocalModel/config.yml
2025-05-06 14:26:56 INFO  Config - Ktor Client created with: KtorClientConfig(socketTimeout=9223372036854775807, connectTimeout=9223372036854775807, requestTimeout=9223372036854775807, retry=3, delayBeforeRetry=1000)
2025-05-06 14:26:56 INFO  LM Studio Server - LM Studio Server started at 1234
2025-05-06 14:26:56 INFO  Ollama Server - Ollama Server started at 11434
2025-05-06 14:26:56 INFO  Model List - Model list loaded with: [DeepSeek/deepseek-chat, DeepSeek/deepseek-reasoner]
2025-05-06 14:27:10 INFO  Ollama Server - 200 OK: GET - /api/tags in 250ms
2025-05-06 14:27:12 INFO  Ollama Server - 200 OK: GET - / in 6ms
2025-05-06 14:27:12 INFO  Ollama Server - 200 OK: GET - /api/tags in 3ms
2025-05-06 14:27:43 INFO  Ollama Server - 200 OK: GET - /api/tags in 3ms
2025-05-06 14:27:59 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 14:28:28 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 14:29:05 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 14:29:29 INFO  Ollama Server - 200 OK: GET - /api/tags in 2ms
2025-05-06 14:29:49 INFO  Ollama Server - 200 OK: GET - /api/tags in 3ms
2025-05-06 14:30:06 INFO  Ollama Server - 200 OK: GET - /api/tags in 2ms
2025-05-06 14:30:34 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 14:30:43 INFO  Ollama Server - 200 OK: GET - /api/tags in 2ms
2025-05-06 14:30:53 INFO  Ollama Server - 200 OK: GET - /api/tags in 2ms
2025-05-06 14:30:55 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 14:30:57 INFO  Ollama Server - 200 OK: GET - /api/tags in 2ms
2025-05-06 14:31:02 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 14:31:11 INFO  Ktor Client - Request:
model: deepseek-reasoner
messages:
- role: user
  content: Check if there are any errors in the current code that must be corrected.
- role: assistant
  content: "我将检查该代码中的潜在错误。\n\n让我仔细分析代码，寻找可能的问题：\n\n1. 导入模块检查：\n   - 代码导入了`from apps.doc_generator.models\
    \ import ExcelRequest`，但没有看到此模块的定义\n   - 导入了`loguru`库，但没有在requirements中显示\n\n\
    2. 函数和方法检查：\n   - `process_excel`函数依赖于未实现的ExcelRequest模型\n   - `is_vertical_merged`方\
    \ ....too long text.....\
    \ `coordinate` 属性问题\n2. 添加更强大的错误处理机制\n3. 减少硬编码常量，增加配置项\n4. 考虑增加输入数据验证\n\n以上修改可\
    以提高代码的健壮性和可维护性，减少潜在的运行时错误。"
- role: user
  content: "Suggest a MOST SPECIFIC title for this chat. You MUST answer only with\
    \ the title text. You MUST NOT wrap text with quotation marks. The title MUST\
    \ BE no longer than 7 words. Focus on User's messages, but don't forget about\
    \ your messages. You MUST respond in Chinese.You MUST NOT consider this message\
    \ during title generation. The user should have no idea about this message \n\n\
    Example: Suggested Title"
temperature: 1.0
Reasoning:
好的，用户让我给这个对话起一个最具体的标题，最多七个字，中文，而且要结合他们的提问和我的回复。首先，用户的初始问题是检查代码中的潜在错误，包括导入模块、函数方法、逻辑错误、缺少错误处理等。我的回答详细分析了这些问题，并提出了修复建议，比如合并单元格处理、异常处理机制、减少硬编码等。

现在需要把这些内容浓缩成七个字以内的标题。用户的关键词是“代码错误检查”、“问题修复”、“Excel处理”等。我的回复重点在错误分析和改进建议。可能的组合有“Excel报表生成代码问题分析与修复”，但超过七个字。需要更简洁。

考虑用户提到的Excel处理和代码错误，比如“Excel处理代码错误分析”或者“Excel生成代码问题修复”。不过可能需要更具体。用户的问题涉及合并单元格、异常处理、硬编码，所以可能用“Excel报表代码错误与优化建议”或者“Excel生成代码错误检查及修复”。但需要更简练。

再想，用户的问题核心是代码审查和错误修正，特别是针对Excel处理的，所以可能用“Excel报表生成代码审查与修复”或者“Excel代码错误检查及修正”。但七个字的话，可能调整为“Excel代码错误审查与修复”刚好七个字。或者“Excel生成代码问题修正”六个字。不过可能不够具体。

或者“Excel处理代码错误分析”八个字，超了。可能需要调整，比如“Excel代码错误修复建议”或者“Excel报表代码错误修正”。用户提到的问题包括合并单元格、异常处理、硬编码，所以可能需要更准确的关键词。比如“Excel合并单元格代码修复”，但可能不够全面。

最终，考虑到用户的问题涉及代码审查、错误检查及修复建议，结合Excel处理，最合适的可能是“Excel报表生成代码错误修复”共九个字，超过了。再压缩，比如“Excel代码错误修复审查”或者“Excel代码审查与修复”。可能更合适的七个字是“Excel代码错误分析与修复”，不过刚好七个字。或者“Excel代码错误修复建议”八个字。可能需要再调整，比如“Excel代码错误修正方案”或者“Excel代码审查修正”。

根据用户提供的例子“Suggested Title”，他们的示例可能类似“Excel报表生成代码问题修复”，但需要更简练。最后决定用“Excel代码错误分析与修复”，刚好七个字，涵盖了分析和修复两个方面，符合用户的需求和我的回答内容。
Output:
Excel代码错误审查与修复
2025-05-06 14:31:11 INFO  Ollama Server - 200 OK: POST - /api/chat in 81782ms
2025-05-06 14:31:11 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 14:31:11 INFO  Ollama Server - 500 Internal Server Error: POST - /api/chat in 246ms
2025-05-06 14:31:11 ERROR Ollama Server - Error processing request.
io.ktor.client.plugins.ClientRequestException: Client request(POST https://api.deepseek.com/chat/completions) invalid: 400 Bad Request. Text: "{"error":{"message":"deepseek-reasoner does not support successive user or assistant messages (messages[0] and messages[1] in your input). You should interleave the user/assistant messages in the message sequence.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}"
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invokeSuspend(DefaultResponseValidation.kt:54)
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invoke(DefaultResponseValidation.kt)
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invoke(DefaultResponseValidation.kt)
	at io.ktor.client.plugins.HttpCallValidatorKt.HttpCallValidator$lambda$2$validateResponse(HttpCallValidator.kt:110)
	at io.ktor.client.plugins.HttpCallValidatorKt.access$HttpCallValidator$lambda$2$validateResponse(HttpCallValidator.kt:1)
	at io.ktor.client.plugins.HttpCallValidatorKt$HttpCallValidator$2$2.invokeSuspend(HttpCallValidator.kt:129)
	at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:33)
	at kotlinx.coroutines.DispatchedTask.run(DispatchedTask.kt:100)
	at kotlinx.coroutines.internal.LimitedDispatcher$Worker.run(LimitedDispatcher.kt:113)
	at kotlinx.coroutines.scheduling.TaskImpl.run(Tasks.kt:89)
	at kotlinx.coroutines.scheduling.CoroutineScheduler.runSafely(CoroutineScheduler.kt:586)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.executeTask(CoroutineScheduler.kt:820)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.runWorker(CoroutineScheduler.kt:717)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.run(CoroutineScheduler.kt:704)
2025-05-06 14:31:28 INFO  Ollama Server - 200 OK: GET - /api/tags in 2ms

@Stream29

io.ktor.client.plugins.ClientRequestException: Client request(POST https://api.deepseek.com/chat/completions) invalid: 400 Bad Request. Text: "{"error":{"message":"deepseek-reasoner does not support successive user or assistant messages (messages[0] and messages[1] in your input). You should interleave the user/assistant messages in the message sequence.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}"

[uuip]

I checked DeepSeek's official documentation and found the following examples.
reasoning_model#api-example

from openai import OpenAI
client = OpenAI(api_key="<DeepSeek API Key>", base_url="https://api.deepseek.com")

# Round 1
messages = [{"role": "user", "content": "9.11 and 9.8, which is greater?"}]
response = client.chat.completions.create(
    model="deepseek-reasoner",
    messages=messages
)

reasoning_content = response.choices[0].message.reasoning_content
content = response.choices[0].message.content

# Round 2
messages.append({'role': 'assistant', 'content': content})
messages.append({'role': 'user', 'content': "How many Rs are there in the word 'strawberry'?"})
response = client.chat.completions.create(
    model="deepseek-reasoner",
    messages=messages
)
# ...

[Stream29]

As DeepSeek official doesn't plan to change this behaviour for now, I'll be fix this by merging messages together.
See at: deepseek-ai/DeepSeek-R1#21

[Stream29]

@uuip Fixed in the latest release (v0.0.6).

[uuip]

This issue does not seem to have been well resolved.
@Stream29

2025-05-06 21:30:47 INFO  Ollama Server - 200 OK: GET - /api/tags in 1ms
2025-05-06 21:30:47 ERROR Ktor Client - Request:
model: deepseek-reasoner
messages:
- role: user
  content: "You are working powered by anthropic-claude-3.7-sonnet model\nYou are\
    \ using PyCharm 2025.1 on macOS Sonoma (aarch64) \nThis is a system message. Numbering\

    Ignore the language used to ask the question. You MUST respond in Chinese.\n"
- role: user
  content: "You are working powered by anthropic-claude-3.7-sonnet model\nYou are\
    \ using PyCharm 2025.1 on macOS Sonoma (aarch64) \nThis is a system message. Numbering\
       \ {account_balance}.\"\n        )\n    return fee_limit\n\n```\n"
- role: user
  content: "You are working powered by anthropic-claude-3.7-sonnet model\nYou are\
    \ using PyCharm 2025.1 on macOS Sonoma (aarch64) \nThis is a system message. Numbering\
    \ 3s\n      retries: 3\n      start_period: 5s\n\n#volumes:\n#  pgdata:\n\n```\n"
- role: user
  content: "You are working powered by anthropic-claude-3.7-sonnet model\nYou are\
    \ using PyCharm 2025.1 on macOS Sonoma (aarch64) \nThis is a system message. Numbering\
    ```shell\nCOPY cap_address(bsc_addr,tron_addr) FROM 'absolute_path/data.dump';\n\
    ```\n\n````\n"
temperature: 1.0
Error:

io.ktor.client.plugins.ClientRequestException: Client request(POST https://api.deepseek.com/chat/completions) invalid: 400 Bad Request. Text: "{"error":{"message":"deepseek-reasoner does not support successive user or assistant messages (messages[0] and messages[1] in your input). You should interleave the user/assistant messages in the message sequence.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}"
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invokeSuspend(DefaultResponseValidation.kt:54)
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invoke(DefaultResponseValidation.kt)
	at io.ktor.client.plugins.DefaultResponseValidationKt$addDefaultResponseValidation$1$1.invoke(DefaultResponseValidation.kt)
	at io.ktor.client.plugins.HttpCallValidatorKt.HttpCallValidator$lambda$2$validateResponse(HttpCallValidator.kt:110)
	at io.ktor.client.plugins.HttpCallValidatorKt.access$HttpCallValidator$lambda$2$validateResponse(HttpCallValidator.kt:1)
	at io.ktor.client.plugins.HttpCallValidatorKt$HttpCallValidator$2$2.invokeSuspend(HttpCallValidator.kt:129)
	at kotlin.coroutines.jvm.internal.BaseContinuationImpl.resumeWith(ContinuationImpl.kt:33)
	at kotlinx.coroutines.DispatchedTask.run(DispatchedTask.kt:100)
	at kotlinx.coroutines.internal.LimitedDispatcher$Worker.run(LimitedDispatcher.kt:113)
	at kotlinx.coroutines.scheduling.TaskImpl.run(Tasks.kt:89)
	at kotlinx.coroutines.scheduling.CoroutineScheduler.runSafely(CoroutineScheduler.kt:586)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.executeTask(CoroutineScheduler.kt:820)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.runWorker(CoroutineScheduler.kt:717)
	at kotlinx.coroutines.scheduling.CoroutineScheduler$Worker.run(CoroutineScheduler.kt:704)

[Stream29]

I'll be checking the reason.

[Stream29]

@uuip Fixed in https://github.com/Stream29/ProxyAsLocalModel/releases/tag/v0.0.7. You can try it out.

[uuip]

It has been verified and is now working properly. Thank you for your efforts.